| acl.default.jobs.leader |
string |
VIEW,EXISTS,CHOWN |
|
Default ACL for FlowTracer jobs assigned to the leader role. |
| alerts.max |
integer |
50 |
[5--1000] |
Maximum number of alerts. If more alerts are created, the oldest with the lowest level is archived. |
| allowcoredump |
boolean |
0 |
0,1 |
If nonzero, request the vovserver to dump core even if it was started from a
shell where limitcoredumpsize was zero. |
| allowForeignJobsOnUserTaskers |
boolean |
0 |
0,1 |
When set to 1, allows foreign users to submit jobs that run on a non-root
tasker that is owned by a different user. When set to 0, foreign user-submitted jobs
will not be scheduled on a non-root tasker. |
| allowUidForSecurityFile |
integer |
0 |
[0---2147483647] |
Additional UID that can own the security.tcl file |
| autoForgetFailed |
time spec |
2 days |
[5m,infinity] |
If the autoForget bit is set on a job, the job will be forgotten
automatically upon failure after the specified time.
|
| autoForgetOthers |
time spec |
2 days |
[5m,infinity] |
If the autoForget bit is set on a job, the job will be forgotten
automatically if it is not scheduled after the specified time.
|
| autoForgetRemoveLogs |
boolean |
0 |
0/1 |
Try to remove the log files associated with auto-forgotten jobs. |
| autoForgetValid |
time spec |
1 hour |
[5m,infinity] |
If the autoForget bit is set on a job, the job will be forgotten
automatically upon successful termination after the specified time.
|
| autoLogout |
time spec |
4 hours |
[10m,infinity] |
The period of inactivity after which the system automatically
logs a user out of the browser interface. The minimum value
for this parameter is 10 minutes.
|
| autoRescheduleCount |
integer |
4 |
0-10 |
Specifies the maximum number of reschedule attempts per job.
|
| autoRescheduleOnNewHost |
boolean |
1 |
0/1 |
Controls whether auto-rescheduling will avoid the same host (default) or just
the tasker, allowing the job to run on a different tasker on the same host. Does not apply to auto-rescheduling based on
exit status (see exit-status docs). |
| autoRescheduleThreshold |
time spec |
0s |
0s-60s |
This parameter deals with jobs that fail quickly. If the failure time is less
or equal to the specified value, the job is automatically resubmitted at high
priority, with the name of the failing host as a negated resource, so that the job
will be placed on a different host (or tasker, configurable) for
the next run. If the autoRescheduleThreshold is 0, automatic resubmission is
disabled. A maximum of 4 auto-reschedules are permitted by default. This maximum can
be controlled via the autoRescheduleCount parameter. Host/tasker
behavior can be controlled via the autoRescheduleOnNewHost parameter. Does not apply
to auto-rescheduling based on exit status (see exit-status docs). |
| autoShutdown |
timespec |
2w |
[15m,infinity] |
A time specification. This parameter controls how long the server must wait
since the last job start before shutting down. If the value is 0, the server never
shuts down. The default value is 2 weeks (2w). The minimum acceptable shutdown
period is 15 minutes. |
| blackholeDiscardTime |
timespec |
10m00s |
[2m00s--10h00m] |
Failed jobs older than this time are discarded in the blackhole computation |
| blackholeFailRate |
double |
0.900000 |
[0.250000--1.000000] |
Fraction of failed jobs that are running well on another tasker |
| blackholeFailedJobs |
integer |
5 |
[2--10000] |
Number of failed jobs on a tasker to consider the tasker a black hole |
| blackholeMaybeTime |
timespec |
10s |
[0s--1h00m] |
Suspension time for tasker that maybe a black hole, but we
are not sure yet |
| blackholeSuspendTime |
timespec |
3m00s |
[1m00s--10h00m] |
Suspension time for a black hole |
| bucketValidTaskerCountPeriod |
timespec |
5m00s |
[0s--41d16h] |
Control frequency of determining if any valid taskers exist
for a bucket.If this parameter is 0, then the facility is turned off. |
| cgi.max |
integer |
40 |
[10--200] |
Max number of concurrent CGI pages that can be generated |
| checkoutHostLowerCase |
boolean |
0 |
0/1 |
If set, host name is always forced to lower case in checkouts |
| checkoutUserLowerCase |
boolean |
0 |
0/1 |
If set, user name is always forced to lower case in checkouts |
| comm.buffer.compress.level |
integer |
1 |
[1--9] |
Compression level between 1 and 9. 1 is speed first and 9 is percentage of string data in buffer to trigger buffer compression |
| comm.buffer.compress.packetSizeThreshold |
integer |
4000 |
[100--1000000] |
Minimum packet size to trigger buffer compression |
| comm.buffer.compress.stringPercentThreshold |
integer |
50 |
[1--100] |
Minimum percentage of string data in buffer to trigger buffer compression |
| comm.maxBufSize |
unsigned |
16777216 |
[65536--2147483647] |
Maximum size of communication buffers between server and client |
| comm.maxTimeToCompleteFlush |
integer |
5000 |
[1000--20000] |
|
| comm.minTimeForCompulsoryFlush |
integer |
1000 |
[1000--2000] |
|
| cpuprogressWindowSize |
integer |
1 |
[1--1440] |
Number of samples to average when calculating cpuprogress (allowed range: 1-1440, default value: 1) |
| crashRecoveryMaxExtension |
timespec |
1m00s |
[0s--30m00s] |
Max extension of crash recovery period due to taskers
reconnecting during quiet time. |
| crashRecoveryPeriod |
time spec |
60s |
[30s,5m] |
This parameter specifies how long the server is in crash recovery mode. The default value is for 60 seconds after which the serer will enter normal operation.
|
| crashRecoveryQuietTime |
timespec |
30s |
[0s--5m00s] |
Possible crash recovery delay due to tasker reconnection
during quiet time before crash recovery ends. |
| dashboard.threshold.bucketcount.critical |
integer |
300 |
[1--1000000000] |
Minimum number of buckets to mark as "critical" bucket count in the web
dashboard. (default: 300) |
| dashboard.threshold.bucketcount.high |
integer |
200 |
[1--1000000000] |
Minimum number of buckets to mark as "high" bucket count in the web
dashboard. (default: 200) |
| dashboard.threshold.bucketcount.low |
integer |
50 |
[1--1000000000] |
Minimum number of buckets to mark as "low" bucket count in the web
dashboard. (default: 100) |
| dashboard.threshold.bucketcount.medium |
integer |
100 |
[1--1000000000] |
Minimum number of buckets to mark as "medium" bucket count in the web
dashboard. (default: 100) |
| dashboard.threshold.schedulertime.critical |
double |
3.000000 |
[0.001000--30.000000] |
Minimum scheduler time to mark as "critical" scheduler activity in the web
dashboard, in seconds. (default: 3.0) |
| dashboard.threshold.schedulertime.high |
double |
1.000000 |
[0.001000--30.000000] |
Minimum scheduler time to mark as "high" scheduler activity in the web
dashboard, in seconds. (default: 1.0) |
| dashboard.threshold.schedulertime.low |
double |
0.100000 |
[0.001000--30.000000] |
Minimum scheduler time to mark as "low" scheduler activity in the web
dashboard, in seconds. (default: 0.5) |
| dashboard.threshold.schedulertime.medium |
double |
0.500000 |
[0.001000--30.000000] |
Minimum scheduler time to mark as "medium" scheduler activity in the web
dashboard, in seconds. (default: 0.5) |
| dashboard.threshold.serversize.critical |
integer |
90 |
[0--100] |
Minimum percentage to mark as "critical" server size in the web
dashboard, as a percentage of total host RAM. (default: 90) |
| dashboard.threshold.serversize.high |
integer |
60 |
[0--100] |
Minimum percentage to mark as "high" server size in the web
dashboard, as a percentage of total host RAM. (default: 60) |
| dashboard.threshold.serversize.low |
integer |
20 |
[0--100] |
Minimum percentage to mark as "low" server size in the web
dashboard, as a percentage of total host RAM. (default: 40) |
| dashboard.threshold.serversize.medium |
integer |
40 |
[0--100] |
Minimum percentage to mark as "medium" server size in the web
dashboard, as a percentage of total host RAM. (default: 40) |
| defaultStopSignalCascade |
string |
TERM,HUP,INT,KILL |
|
Default stop signal cascade; used for job control |
| defaultStopSignalDelay |
integer |
1 |
[0--20] |
Default stop signal delay; number of seconds to wait between transmitting
signals to taskers when using vovstop and related commands.
(allowed range: 0-20, default: 1 |
| defaultSuspendSignalCascade |
string |
STOP |
|
Default suspend signal; used for job control |
| disablefileaccess |
boolean |
0 |
|
If this parameter is greater than 0 (zero), the server will not access
any file that is not under $VOVDIR or $VNCSWD.
By 'access' we mean any operation on the file like stat(), open(), access().
This is a protection to
prevent a possible server freeze if
it tries to access files on an dead NFS server.
If this parameter is greater than 1 (one), then not only the server
disables access to files, but also the CGI clients that could possibly
show file information will disable such functionality.
In summary:
- Under normal conditions, set this parameter to 0
- If you are concerned about the vovserver blocking on a bad NFS mount, set this parameter to 1
- If you want maximum information security, set this parameter to 2.
|
| diskspacecheck.minfreemb |
integer |
1000 |
[1--50000] |
Specify free disk space alert generation threshold (in MB) for free disk space checking (allowed range: 1-50000, default: 1000 |
| diskspacecheck.minfreepercent |
integer |
1 |
[1--50] |
Specify free disk space alert generation threshold (in %) for free disk space checking (allowed range: 1-50, default: 1 |
| emptyBucketAge |
integer |
120 |
[0--86400] |
Delete empty buckets older than the specified age in seconds. |
| enableLdap |
boolean |
0 |
0/1 |
If set, the system uses also LDAP for browser authentication.
Available on the following architectures: macosx, linux64, armv8
|
| enablePam |
boolean |
1 |
0/1 |
If set, the system uses PAM for browser authentication.
|
| enableTimestampCheckAllFiles |
boolean |
1 |
0/1 |
On large traces, it is good to completely disable the checking of the timestamps of files |
| enableTimestampCheckForRetrace |
boolean |
1 |
0/1 |
On large traces, it is good to disable the checking of the timestamps of the upcone of a target because it may take too long |
| enableWaitReasons |
boolean |
1 |
0/1 |
Enable logging of Wait-Reasons. |
| enterpriselicense.burst |
boolean |
0 |
0/1 |
Use burst licenses if available |
| enterpriselicense.delay.decrease |
timespec |
5m00s |
[10s--1h00m] |
With enterprise licensing in automatic mode, wait this much before decreasing again |
| enterpriselicense.delay.increase |
timespec |
4s |
[1s--5m00s] |
With enterprise licensing in automatic mode, wait this much before increasing again |
| epollbuf |
integer |
50 |
[50--1024] |
Epoll buffer size |
| failover.maxdelaytovote |
integer |
120 |
[10--3600] |
Specifies the maximum duration during which a vovtasker can
participate in a failover server election (default: 120). |
| failover.usefailovertaskergrouponly |
boolean |
0 |
0/1 |
If set, only taskers in the failover tasker group participate in the failover server election (default: 0). |
| fairshare.allowAdminBypass |
boolean |
0 |
0/1 |
Allow ADMIN to bypass ACL when modifying a FairShare Group |
| fairshare.default.weight |
integer |
100 |
[0--100000] |
Default weight assigned to new FairShare groups. Can also be controlled by a sibling group called 'default'. |
| fairshare.default.window |
timespec |
1h00m |
[0s--7d00h] |
Default window assigned to new FairShare groups. Normally the groups inherit the window from their parent of from a sibling group called 'default'. |
| fairshare.maxjobsperbucket |
integer |
20 |
[1--1000000] |
Max number of jobs to dispatch from each bucket at once. |
| fairshare.maxjobsperloop |
integer |
20 |
[1--1000000] |
Max number of jobs to dispatch on each loop, after which the FairShare stats will be recomputed. |
| fairshaire.overshoot.damping |
boolean |
1 |
0/1 |
If set, each bucket will be limited by its max dispatch depth for each
scheduler cycle (default: 1). |
| fairshare.relative |
integer |
1 |
[0--2] |
Assuming: t=target allocation; h=historic allocation in window; r=running allocation now.
If set to 2, compute fairshare as a weighted sum of d = (r-t) + alpha(h-t). If set to 1, compute fairshare distance relative to the fairshare target d=((r-t)+(h-t))/t (of course assuming t>0). If set to 0, compute fairshare distance by a simple difference between actual and target d=(r-t)+(h-t). |
| fairshare.relative_alpha |
double |
10.000000 |
[0.001000--1000.000000] |
Weight of historic distance relative to running distance in computing value for fairshare ranking. Only used if fairshare.relative is set to 2. |
| fairshare.updatePeriod |
timespec |
0s |
[0s--1h00m] |
Update FairShare stats no more frequently than specified period |
| hog.protection.clientdelay |
integer |
1 |
[1--600] |
Specify the client processing delay (in seconds) for users under hog protection (allowed range: 1-600, default: 1 |
| hog.protection.enable |
boolean |
0 |
0/1 |
Enable hog protection in scheduler (default: 0) |
| hog.protection.jobcountthreshold |
integer |
100000 |
[1000--9999999] |
Specify currently-managed (all states) job count threshold to trigger hog protection against a user (allowed range: 1000-9999999, default: 100000 |
| http.proxytimeout |
unsigned |
3600 |
[5--10000] |
HTTP Proxy Timeout (seconds) |
| httpSecure |
boolean |
1 |
|
This should be set to 1 or "enable" if you use the browser user interface. If
it is set to "0", browser users do not have to authenticate, and hence the server
does not know their login identity. The default value is 1. When this parameter is
1, the server requires basic authentication to serve most of its pages. By setting
this parameter to 0 (zero), any access via the HTTP protocol is allowed, but the
server will not know user's login identity. This is provided for backwards
compatibility, but the use of the value is discouraged. If this parameter is set to
2, or "disable", or any other number, then the HTTP interface is completely
disabled, |
| http.workerthreads |
integer |
n/a |
n/a |
Obsolete: no longer used. |
| idrange |
range |
100,000,000 |
1,000,000 - 2,000,000,000 |
The range to be used for VovIds. This is also related to vovIdLow and
vovIdHigh. The typical range is 2000-99999999 |
| jetstreams.debug |
boolean |
0 |
0/1 |
Enable jetstream debug messaging in scheduler (default: 0) |
| jetstreams.enable |
boolean |
1 |
0/1 |
Enable jetstreams in scheduler (default: 1) |
| jetstreams.threshold |
integer |
5 |
[1--86400] |
Specify maximum job duration (in seconds) for utilizing jetstreams (allowed range: 1-86400, default: 5 |
| la.adjustForUnconsumed.enable |
boolean |
1 |
0/1 |
Enable/disable adjust for unconsumed allocations. |
| la.adjustForUnconsumed.timeWindow |
timespec |
2m00s |
[1m00s--5m00s] |
Time window to check for unconsumed allocations for adjustment. |
| la.feature.expirationThreshold |
timespec |
1h00m |
[30s--1d00h] |
Time threshold since feature data last received from LM, after which LA will stop allocating this expired feature. |
| la.ooq.waves.disable |
boolean |
0 |
0/1 |
Disable wave based OOQ estimation. |
| liverecorder.logdir |
string |
. |
|
Directory in which the Live Recorder recording file should be saved. The directory must exist. Default is '.', which is the CWD of the process that is running Live Recorder. |
| liverecorder.logsize |
integer |
256 |
[256--65536] |
Live Recorder log size in MB (range: 256-65536 default: 256) |
| log.nodestatus.changedetail |
boolean |
0 |
0/1 |
Enable additional detail on node status changes |
| log.nodestatus.loggingtype |
integer |
0 |
[0--3] |
Global nodestatus logging type (0=NONE, 1=ALL, 2=INVALID, 3=DOWNCONE) |
| logs,checkouts,compress |
boolean |
1 |
|
You can control which log files get compressed by editing the
policy.tcl file and setting the config(log,XXXX,compress)
flags. |
| logs,checkouts,handler |
string |
|
|
A command line to process the log file as it is being created. The command line
should contain the variables @FILENAME@ and @EXPIRE@. |
| logs,jobs,compress |
boolean |
|
|
You can control which log files get compressed by editing the
policy.tcl file and setting the config(log,XXXX,compress)
flags. |
| logs,jobs,handler |
string |
|
|
A command line to process the log file as it is being created. The command line
should contain the variables @FILENAME@ and @EXPIRE@. |
| logs,journals,compress |
boolean |
0 |
|
You can control which log files get compressed by editing the
policy.tcl file and setting the config(log,XXXX,compress)
flags. |
| logs,journals,handler |
string |
|
|
A command line to process the log file as it is being created. The command line
should contain the variables @FILENAME@ and @EXPIRE@. |
| logs,resources,compress |
boolean |
0 |
|
You can control which log files get compressed by editing the
policy.tcl file and setting the config(log,XXXX,compress)
flags. |
| logs,resources,handler |
string |
|
|
A command line to process the log file as it is being created. The command line
should contain the variables @FILENAME@ and @EXPIRE@. |
| logs,server,compress |
boolean |
0 |
|
You can control which log files get compressed by editing the
policy.tcl file and setting the config(log,XXXX,compress)
flags. |
| logs,server,handler |
string |
|
|
A command line to process the log file as it is being created. The command line
should contain the variables @FILENAME@ and @EXPIRE@. |
| logs,tasker,compress |
boolen |
1 |
1/0 |
You can control which log files get compressed by editing the
policy.tcl file and setting the config(log,XXXX,compress)
flags. |
| logs,waitreason,compress |
boolean |
0 |
|
You can control which log files get compressed by editing the
policy.tcl file and setting the config(log,XXXX,compress)
flags. |
| logs,waitreason,handler |
string |
|
|
A command line to process the log file as it is being created. The command line
should contain the variables @FILENAME@ and @EXPIRE@. |
| longServiceWarningDuration |
double |
0.800000 |
[0.000000--1000000.000000] |
Warn if a service takes a long time, as defined by this parameter (in seconds). |
| maxAgeRecentJobs |
timespec |
60 |
[0,3600] |
This defines the max age of the jobs in the set System:transitionsRecent, which
contains all the jobs that have recently finished. Due to various scheduling delays,
it is possible for some jobs in the set to be somewhat older.
|
| maxBufferSize |
unsigned |
16777216 |
[65536--2147483647] |
Maximum size of communication buffers between server and client. The same with comm.maxBufSize |
| maxChildren |
integer |
20 |
|
Maximum number of children managed by vovserver. This mostly affects the number
of concurrent CGI requests that the server can manage. |
| maxCommandLineLength |
integer |
40960 |
[128,100000] |
Number of characters in the longest command line that
will be accepted by the system.
A job containing a longer command line will be rejected.
|
| maxEnvLength |
integer |
512 |
[128,16000] |
Maximum length allowed for the environment field in a job.
Example: If the environment is BASE+GNU, length of the string is 8.
|
| maxJobArray |
integer |
10000 |
[10,100000] |
This parameter defines the maximum size of a job array.
|
| maxLevel |
integer |
4095 |
[3,4095] |
Maximum number of levels in the dependency graph. Dependency declarations that
cause the levels in the graph to exceed this value are rejected. |
| maxNormalClients |
integer |
400 |
|
Maximum number of normal clients per user. This is
a protection against denial-of-service attacks by a single user.
|
| maxNotifyBufferSize |
integer |
1000000 |
[100k,40M] |
In bytes, this parameter controls the size of the buffers used
internally to notify asynchronous clients. If you see "overflow"
events and are annoyed by it, you can increase the size of the
buffer.
|
| maxNotifyClients |
integer |
40 |
[0-100000] |
Maximum number of notify clients per user. This is
a protection against denial-of-service attacks by a single user.
The 'notify' clients are those that tap into the
event stream. They are used for example in waiting for
jobs to finish and to update the GUI.
|
| maxORsInResourceExpressions |
integer |
20 |
[0--25] |
Max number of OR operators in a resource expression |
| maxPathLength |
integer |
1024 |
[128,16000] |
Maximum length of names of files, including the end of line terminator.
This is different from all other "max" parameters for jobs.
This means that if this parameter has value 1024, then the actual max length of a path is actually 1024-1=1023.
|
| maxResMap |
integer |
5000 |
[1, infinity] |
Maximum resources map number that can be set in the resource map set. |
| maxResourcesLength |
integer |
1024 |
[128,16000] |
Maximum length of the resources field for a job. |
| maxSmartSets |
integer |
50 |
[10--5000] |
Max number of smart sets |
| maxeventqueue |
integer |
100000000 |
[1000--500000000] |
Max number of events in event queue |
| metrics.autoSavePeriod |
timespec |
5m00s |
[1m00s--1h00m] |
Period for auto-saving metrics data files to disk. |
| metrics.enable |
integer |
7 |
[0--7] |
Enable the metrics system. The old name 'enableMetrics' can also be used. The value is a mask, with the following bit values: 0=disable_all_metrics 1=enable_server_metrics 2=enable_design_metrics 4=enable_fairshare_metrics 7=enable_all_metrics |
| metrics.keepFilesPeriod |
timespec |
30d00h |
[0s--3y00d] |
How long to keep saved metrics data files. |
| metrics.maxmem |
integer |
2000 |
[100--100000] |
Max size in MB for each metrics family. If the memory exceeds this amount, the system drops some metrics or some points |
| minMemFreeOnServer |
integer |
10 |
[0--10000] |
Issue alert if free memory on server goes below this minimum threshold, in MB |
| minSwapFreeOnServer |
integer |
10 |
[0--10000] |
Issue alert if free swap on server goes below this minimum threshold, in MB |
| mm.fork |
boolean |
0 |
0/1 |
Enable use of MADV_DONTFORK for client buffers |
| mm.huge |
boolean |
0 |
0/1 |
Enable Huge Pages |
| nfsdelay |
timespec |
0s |
[0s--5m00s] |
If this delay is set, the vovserver will not check the timestamps of files
that are younger than this delay, because of the uncertainty about the status of the
NFS cache. If the NFS server and the VOV server are running on the same host, you
can set this delay to zero. For tight edit-compile loops, you are advised to set
this delay to zero and to do the editing of the primary inputs on the same machine
that runs the VOV server, so that there is not NFS cache effect. |
| notifyMaxEffort |
integer |
20 |
[5--50] |
Percent of time to give to notification of clients |
| notifySkip |
integer |
0 |
[0--100000] |
OBSOLETE: NO LONGER USED In inner loop, do not notify clients for each event, but skip a few. |
| preemption.load |
boolean |
1 |
0/1 |
load/unload preemption module |
| preemption.log.allrules |
boolean |
0 |
0/1 |
Enable debug messages for all preempt rules |
| preemption.log.verbosity |
integer |
0 |
[0--10] |
Control verbosity of log messages for preemption. |
| preemption.max.time.overall |
double |
0.300000 |
[0.001000--10.000000] |
Max time spent in preemption for each loop |
| preemption.max.time.rule |
double |
0.300000 |
[0.001000--10.000000] |
Max time spent in preemption for each rule. Any rule that exceeds this will be disabled. |
| preemption.module |
string |
libpreemption.so |
|
preemption module name |
| preemption.rule.cooldown |
time spec |
"1y" |
[0, "1y"] |
Specifies how long rules are disabled due to exceeding
preemption.max.time.rule. 1y = Indefinite, 0 = generate alert only. Default is 1y
(Indefinite). |
| preemptionFairshare |
timestamp |
0 |
|
If greater than the current date, enable computation of statistics for FairShare of preemption. Normally set by
vovpreemptd. For testing, use
vtk_server_config "preemptionFairshare" "on", which activates
the computation for the next 10 minutes. |
| preemptionPeriod |
time spec |
3s |
[0,1000h] |
This parameter specifies how often Accelerator will attempt to perform
preemption. The default value is 3 seconds which will cause preemption to be
attempted every 3 seconds. If a value of 0 is specified then this will disable
preemption. |
| preemptionResourceRelease |
time spec |
90s |
[0,1000h] |
This parameter controls the frequency of releasing features for suspended jobs as a result of preemption. If this parameter is 0, the code is never called.
|
| processRetentionTime |
timespec |
5s |
[0s--5m00s] |
How long a process record (see procinfo feature) is kept in memory since the last update. |
| prop.maxNameSize |
unsigned |
255 |
[16--1024] |
Max length of name a property |
| prop.maxStringSize |
unsigned |
131071 |
[128--1048576] |
Max length of value of a string property. Any value longer than this will be truncated. |
| prSaveTimeout |
timespec |
1h |
[60s--41d16h] |
Time (seconds) after which an incomplete PR save operation exits. |
| readonlyport |
integer |
0 |
[0--65535] |
The readonly port for vovserver |
| rejectIndirectTaskerJobs |
boolean |
0 |
0/1 |
Reject jobs submitted through the binary indirect tasker |
| resMapGroupDefault |
integer |
50 |
-1 (unlimited) |
This parameter sets the default value for the resource map "Group:groupname"
for a new group. This resource map limits the maximum number of jobs that the group
can run at the same time. For each individual group, you can control the value of
the map with a vtk_resourcemap_set Group:groupname number statement
in the resources.tcl file. A negative value of this parameter
is interpreted as "unlimited",. |
| resMapToolDefault |
integer |
|
-1 (unlimited) |
This parameter sets the default value for the resource map "Tool:toolname" for
a new tool. This resource map limits concurrent number of jobs using the tool. For
each individual tool, you can control the value of the map with a
vtk_resourcemap_set Tool:toolname number statement in the
resources.tcl file. A negative value of this parameter is
interpreted as "unlimited". |
| resMapUserDefault |
integer |
50 |
-1 (unlimited) |
This parameter sets the default value for the resource map "User:username" for
a new user. This resource map limits the maximum number of jobs that the user can
run at the same time. For each individual user, you can control the value of the map
with a vtk_resourcemap_set User:username number statement in the
resources.tcl file. A negative value of this parameter is
interpreted as "unlimited". |
| resmap.max.map.length |
integer |
1024 |
[512--64000] |
Max length of the map field in a resourcemap |
| resmap.max.maps.in.expression |
integer |
300 |
10, 50000 |
Max number of resourcemaps considered in the evaluation of a resource expression.
This is a protection against cycles in the resource maps (e.g. a ←b b←c c←a).
The typical value is less or equal to the total number of resource maps in the system.
|
| resmap.max.nru.time |
double |
0.100000 |
[0.000100--1.000000] |
Max time allowed to compute NRU handles |
| resmap.sw.types |
string |
License Limit Policy User Group Priority |
|
Comma or Space separated list of types of resources that are guaranteed to be resource maps (i.e. SW resources), even if they are not currently defined as a resource map. This parameter is typically an empty list in WX and includes 'License Limit Policy ...' for the other products. |
| resuserDisableMatchingThreshold |
integer |
1000 |
[0--1000000] |
Threshold of LM handles and FT jobs to match. If the sum of LM handles and FT jobs exceeds this threshold, do not perform matching. |
| resuserEnableNRUMatches |
boolean |
1 |
0/1 |
Matching check should include checking for Not-Requested-Used licenses. |
| resusermatchtolerance |
timespec |
0s |
[0s--10m00s] |
Tolerance (in seconds) allowed when matching jobs and handles |
| resusermaxmatches |
integer |
6 |
[0--40] |
Maximum number of 'also' matches between jobs and lm handles |
| runAsWX |
integer |
0 |
[0--1] |
Set up this project to run as WX. |
| sanity,smartsets |
boolean |
0 |
0/1 |
enable sanity check of smart sets |
| saveToDiskPeriod |
time spec |
2 hours |
[10m,infinity] |
This parameter controls the period of recurrent saves to disk of all the trace
information. If the trace is very large (>400MB) the recurrent save may take a long
time. The minimum value for this parameter is 10 minutes. The default value is 2
hours. A value of zero or less disables the periodic saving. |
| sched.busy.thresh |
integer |
3 |
[2--1000000] |
If the delay between poll() is more than this threshold, then the server is considered to be 'busy' |
| sched.maxpostponedjobs |
integer |
10000 |
[1--1000000] |
Exit dispatch loop after this number of jobs have been postponed, indicating a loaded or busy farm |
| sched.megapoll.threshold |
integer |
100 |
[5--10000000] |
Threshold that defines how many clients are needed to declare that we are in a 'mega-poll' situation |
| sched.taskerstats.enable |
boolean |
0 |
0/1 |
Enable scheduler statistics for taskers (used in
unusedhw.cgi) |
| schedMaxEffort |
integer |
20 |
5 50 |
If the scheduler becomes expensive, i.e. if it takes more than 0.02 seconds
to dispatch jobs, the vovserver
adaptively skips some scheduling cycle to give time
to the clients to interact with the vovserver.
In this way, the scheduler effectively changes from a pure event-driven
behavior to a cycle-based behavior.
The cycle is controlled by the schedMaxEffort variable, which sets the
ratio between the effort spent in the scheduler and the effort
spend elsewhere.
Example: If the schedMaxEffort is 20 (meaning 20% of time should
be spent in the scheduler)
and the scheduler time is 0.1s, then the scheduler will be called
no more frequently than once every 0.5 seconds (0.1*100/20=0.5)
|
| sds.enabled |
boolean |
0 |
0/1 |
enable/disable Streaming Data Service |
| seatlic_max |
integer |
2147483647 |
[0--2147483647] |
Override the maximum number of seat licenses that can be checked out (default:
unlimited |
| tasker.childProcessCleanup |
boolean |
0 |
0/1 |
If TRUE, taskers should kill all of a job's child processes
when that job exits (i.e. assume child processes running after jobs exit are zombie
processes); default is FALSE. |
| tasker.max.reserve |
integer |
10 |
[0--100] |
Maximum number of reservations per tasker |
| tasker.maxWaitToReconnect |
timespec |
3m00s |
[5s--1y00d] |
Time to wait after losing connection to server before reconnecting again. |
| tasker.props.enable |
boolean |
0 |
0/1 |
Enable saving of message properties for all taskers |
| slave.uninterruptableSignalCascades |
boolean |
0 |
0/1 |
if TRUE, slaves will ignore incoming job control requests if an existing request
is in-process; default is FALSE. |
| taskerheartbeat |
integer |
60 |
[1--300] |
Default heartbeat for taskers, in seconds. Can be overridden
for individual taskers via vovtaskermgr configure or
by starting the vovtaskers with the -U argument |
| ssl.enable |
boolean |
0 |
0/1 |
Enable SSL for rest web server |
| ssl.enableRawUrlOnHttp |
boolean |
1 |
0/1 |
Enable /raw url's for SSL with REST server |
| substituteArrayIdInResources |
boolean |
0 |
0/1 |
Substitute references to @ARRAYID@ in resource string with array ID (default: 0). Caution should be used because this will result in a job queue bucket being created for each submission of a job array. |
| substituteArrayIndexInResources |
boolean |
0 |
0/1 |
Substitute references to @INDEX@ in resource string with array index (default: 0). Caution should be used because this will result in a job queue bucket being created for each job in an array. |
| substituteJobIdInResources |
boolean |
0 |
0/1 |
Substitute references to @JOBID@ and @IDINT@ in resource string with job ID (default: 0). Caution should be used because this will result in a job queue bucket being created for each job. |
| switchToExtBufThresh |
integer |
100000 |
[10k,1...31] |
In bytes, this parameter controls the size of the internal buffer used for vtk_set_list_elements_enhanced.
If the size exceeds this parameter, then vovserver switches to an external buffer.
|
| tasksMaxEffort |
integer |
20 |
[5--50] |
Percent of time to give to maintenance tasks |
| thread.preempt.max |
integer |
40 |
[10--200] |
Max number of concurrent threads for auxiliary preemption work like vovlmremove |
| thread.service.enable.fsgroup |
boolean |
1 |
0/1 |
Enable safe threads for VovFsGroups |
| thread.service.enable.query |
boolean |
1 |
0/1 |
Enable safe threads for queries |
| thread.service.fork.ref.time |
double |
0.020000 |
[0.001000--2.000000] |
Reference cost for a fork() used to decide when to fork() for service threads |
| thread.service.max |
integer |
0 |
[0--50] |
Max number of concurrent forked threads for read-only services |
| thread.service.min.nodes |
integer |
50000 |
[0--40000000] |
Min Number of nodes in set to activate safe threads |
| thread.service.min.resmaps |
integer |
500 |
[0--1000000] |
Min Number of resmaps to activate safe threads |
| thread.service.min.slaves |
integer |
500 |
[0--100000] |
Min Number of taskers to activate safe threads |
| timeTolerance |
time spec |
0 |
[0,600] |
Control the time tolerance of the server.
The default value of 0 may be too strict for some networks. You will
usually need to have ntp running to synchronize the clocks to use zero,
but this is the best practice. Values up to about 3 seconds will not
impose much performance penalty.
|
| trustHostReportedByClient |
boolean |
0 |
0/1 |
Used mostly for demos |
| useepoll |
boolean |
0 |
0/1 |
Switch between poll(0, default) or epoll(1) for client servicing (Linux only). |
| userGroupLowerCase |
boolean |
0 |
0/1 |
Force user group name to be lower case |
| userLowerCase |
boolean |
0 |
|
If set to 1, force all user names to lowercase.
This applies also to ACL and to the security modules (vtk_security).
|
| usrtmp |
path |
/tmp |
|
/tmp, /usr/tmp, /var/tmp, c:/temp |
| vovIdHigh |
integer |
999,999,999 |
[1,000,000 - 2,000,000,000] |
When recycling vovid's, this is the largest usable id. This will be at least
1,000,000 more than vovIdLow. |
| vovIdLow |
integer |
5,000 |
[5,000,10,000,000] |
When recycling vovid's, this is the id which the id generator wraps back
to. |
| vovstring.maxLength |
unsigned |
2147483647 |
[65535--4294967295] |
Max length of a VovString object |
| vovwxd.localresources |
boolean |
0 |
0/1 |
If set, the user facing queue will manage local resourcemaps |
| webport |
integer |
0 |
[0--65535] |
The web port used by internal "rest" http server (possibly with ssl) |
| wx.dispwithbucketres |
boolean |
1 |
0/1 |
Use Bucket:XXX resource shortcut in WX |