On 01/08/14 20:17, Crowe, Tom wrote:
All,

We have been working with robin hood 2.4.3 for 2-3 months now, and we are not getting the numbers that we expected from the rbh-reports. Our filesystem has about 1.4 TB of data in flight and we have never gotten robin hood to report on more than a few hundred TB.

We are running Lustre 2.1.6, and have RBH 2.4.3 installed from RPM's (tmpfs and adm).

We have enabled the change log processing from an admin client (no root squash) and that seems to be parsing the changlog correctly. The filesystem was populated with data before robin hood was installed. So we have been trying to run a "catch up" scan from the same admin client (while the change log reader is processing). Please note this admin client is NOT running on the MDS and is running lustre 2.1.3
Indeed, this is the recommended setup: robinhood is to run on a client, not on the MDS.

The issues we have experienced are mainly hung scans, hung clients (mount hangs up on admin client), and ultimately, incomplete/incorrect data in the RBH database.

Is it "OK" to run the change log reader AND the scan from the same admin client?
Yes, it is supported to run both at the same time.
However, I suggest you first complete the initial scan (while your changelogs keep stacking), and then process the changelog once the scan is completed,
this way you will speedup the scan by offloading the DB.


Is there anything we can do to assist the scan in completing correctly?
Your issues sound like lustre client issues. When the client is hung, dmesg sometimes show the current stuck process and dumps its stack.
Else, try to dump your node and see where it was stuck.
Also see the previous discussion in the mailing list "Slab memory usage during initial robinhood scan" to check if you don't have a similar problem.

If you have no hope to fix the client hangs, you can consider splitting the scan using '--partial-scan' option (rbh 2.4) so you can progressively complete the scan step-by-step, and only have to restart the current and remaining parts if your client hangs.

Regards,
Thomas


Thanks for your assistance.

-Tom


------------------------------------------------------------------------------
Rapidly troubleshoot problems before they affect your business. Most IT
organizations don't have a clear picture of how application performance
affects their revenue. With AppDynamics, you get 100% visibility into your
Java,.NET, & PHP application. Start your 15-day FREE TRIAL of AppDynamics Pro!
http://pubads.g.doubleclick.net/gampad/clk?id=84349831&iu=/4140/ostg.clktrk


_______________________________________________
robinhood-support mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/robinhood-support

------------------------------------------------------------------------------
CenturyLink Cloud: The Leader in Enterprise Cloud Services.
Learn Why More Businesses Are Choosing CenturyLink Cloud For
Critical Workloads, Development Environments & Everything In Between.
Get a Quote or Start a Free Trial Today. 
http://pubads.g.doubleclick.net/gampad/clk?id=119420431&iu=/4140/ostg.clktrk
_______________________________________________
robinhood-support mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/robinhood-support

Reply via email to