Hide Forgot
Description of problem: For gluster volume top read-perf/write-perf when the value of bs*count is large, it blocks glusterd. Version-Release number of selected component (if applicable): master How reproducible: always Steps to Reproduce: 1. run volume top read-perf with large value of bs and count (bs*count should be alrge) Actual results: glusterd gets blocked and return after sometime with non-zero exit status. Now the subsequent gluster operations will fail. Expected results: glusterd should not get blocked. There should either be a limit on the bs*count value or the task should be performed in a different thread. Additional info: This actually is a duplicate of bug https://bugzilla.redhat.com/show_bug.cgi?id=GLUSTER-2720. A patch was sent to this but that doesn't solve the issue. I couldn't reopen that bug so opening a new bug.
CHANGE: http://review.gluster.com/2761 (cli, glusterfsd: Fixes for "volume top {read,write}-perf") merged in master by Vijay Bellur (vijay)
Now there is limit on the size of "bs * count" value which is 10GB. So write-perf won't block glusterd. [root@QA-29 vishwa]# gluster v top hosdu write-perf bs 2789473574385746754678576 count 100 list-cnt 10 'bs * count' value 429496729500 is greater than maximum allowed value of 10GB Usage: volume top <VOLNAME> {[open|read|write|opendir|readdir [nfs]] |[read-perf|write-perf [nfs|{bs <size> count <count>}]]} [brick <brick>] [list-cnt <count>]