Highlighted
2 Bronze

Re: How can I reduce the size of isi_gather_info?

Jump to solution

Looking at our SyncIQ policies, the following 2 settings seem to be the key:

Report Max Age: 4W2D
Report Max Count: 2000

So, the age policy isn't overriding the count policy, and we end up with 2000 SyncIQ reports per policy, e.g.:

# ls /ifs/.ifsvar/modules/tsm/sched/reports/0007430a77cc3fc78755150883d43aa1|wc -l
2010

# du -sh /ifs/.ifsvar/modules/tsm/sched/reports/0007430a77cc3fc78755150883d43aa1
190M /ifs/.ifsvar/modules/tsm/sched/reports/0007430a77cc3fc78755150883d43aa1

 

0 Kudos
Highlighted
Moderator
Moderator

Re: How can I reduce the size of isi_gather_info?

Jump to solution

Hello JohnB999,

There are a few options for reducing the size of log gathers. We detail them below.
 

Option 1: Remove common sources of extra data when gathering logs

OneFS 7.1.1.2 offers the ability to exclude node reboot cores, process cores, and hangdumps when running isi gather info.  When running isi_gather_info, add one or both of these switches to the command to limit what is gathered:
 

--no-dumps--no-cores

Option 2: Remove common sources of extra data from the cluster permanently with isi_gather_info

Instruct isi_gather_info to remove cores and hangdumps from the cluster after they have been gathered

OneFS 7.1.1.2 offers the ability to clean node reboot cores, process cores, and hangdumps after they have been gathered with isi_gather_info.  When running isi_gather_info, add this switch to the command to clean those cores and hangdumps from the cluster:

--clean-all

If you wish to clean only cores, or only hangdumps, you can add one of the following switches to the isi_gather_info command:

--clean-dumps
--clean-cores

Option 3: Manually remove sources of data

vmcore files in /var/crash from node reboot events

When a node reboots, it creates a file in /var/crash with a filename that starts with vmcore.  If those vmcore files have already been uploaded to EMC support for analysis, you can remove the file(s) with this command:
 

isi_for_array "rm -rf /var/crash/vmcore*"

hangdump files in /var/crash from cluster hangdump events

During file locking activity, the isi_hangdump daemon will sometimes trigger a hangdump on all nodes in the cluster which creates a diagnostic file in /var/crash with a filename that starts with isi_hangdump.  If those hangdump files have already been uploaded to EMC support for analysis, you can remove the file(s) with this command:

 isi_for_array "rm -rf /var/crash/isi_hangdump*"

Process core files in /var/crash

When a process terminates abnormally, it usually writes out the contents of its memory into a core file and stores that file in /var/crash with a filename that ends with .core.gz.  If those core files have already been uploaded to EMC support for analysis, you can remove the file(s) with this command:
 

isi_for_array "rm -rf /var/crash/*.core.gz"

rst* logfiles in /ifs/.ifsvar/tmp from previous isi_vol_copy or isi_vol_copy_vnx migrations

During isi_vopy_copy or isi_vol_copy_vnx migrations, those processes create files in /ifs/.ifsvar/tmp that begin with the filename rst.  If you do not need those files any longer, you can remove them from that directory with this command:
 

rm -rf /ifs/.ifsvar/tmp/rst*

An issue in 7.1.0.0 which gathers FSAnalyze reports databases (fixed in 7.1.0.1)

OneFS 7.1.0.0 had a problem with its isi_gather_info script which would cause the FSAnalyze reports databases to be collected, driving up the size of the log gather.  This issue was resolved in OneFS 7.1.0.1.

 

SyncIQ report files

SyncIQ report files can take up larger than expected amounts of space and possibly increase the amount of time it takes isi_gather_info to complete its log gather.  You can adjust the amount of SyncIQ report retention to keep only a certain number of reports, or limit how long it will keep reports via these command line options.  Choose which command suits your needs best and apply it to existing policies.
 

isi sync policies modify <policy> --report-max-age <duration>isi sync policies modify <policy> --report-max-count <count>

Once you have made this change to your policies you can change the defaults for new policies with the following commands:

isi sync settings modify --report-max-age <duration>

isi sync settings modify --report-max-count <count>

Please let us know if you have any other questions.

DELL-Sam L
Dell | Social Outreach Services - Enterprise
Download the Dell Quick Resource Locator app today to access PowerEdge support content on your mobile device! (iOS, Android, Windows)

0 Kudos