Description of problem: I have created 111 VMS on 1 host and turned them on. After the VMS where up, the VDSM process took 300 MB of RAM. Now, 6 day's later, with the system idle - no actions are done on the engine or the VMS, the VDSM takes ~630 GB of RAM as seen below : PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND 18100 vdsm 0 -20 9980.2m 634340 11260 S 14.1 1.0 379:42.58 vdsm Version-Release number of selected component (if applicable): VDSM = 4.18.11 How reproducible: 100% Steps to Reproduce: 1.create 111 VDSM 2.measure the VDSM memory after 6 day's Actual results: VDSM memory doubled it's size in 6 days time. Expected results: after VMS are up, VDSM memory should stay the same. Additional info: No errors are seen at the VDSM logs.
we know for sure we have more usage of the warnings module, so we could have more instances, albeit less serious, of the problem which caused https://bugzilla.redhat.com/show_bug.cgi?id=1364149 Let's start from there.
AFAIU this is not as serious as bug 1364149
yes, the memory leak trend is not as bad as it was on bug 1364149
do we want any more patches?
Tested latest 4.0.5 system, VDSM memory started at 360 MB, checked the system after 10 day's and VDSM is stable around 390 MB, thus Bug is verified.