Message ID | 20210615121517.27517-1-krzysztof.kozlowski@canonical.com |
---|---|
State | New |
Headers | show |
Series | [ACT] UBUNTU: SAUCE: ubuntu_lttng_smoke_test: adjust memory limit to fix OOM | expand |
On 15/06/2021 14:15, Krzysztof Kozlowski wrote: > The test was failing on Azure Standard_B1ms instance due to OOM killer: > > utils:0153| [stdout] gcc: fatal error: Killed signal terminated program cc1 > > In journalctl: > > kernel: cc1 invoked oom-killer: gfp_mask=0x100cca(GFP_HIGHUSER_MOVABLE), order=0, oom_score_adj=0 > ... > kernel: Free swap = 0kB > kernel: Total swap = 0kB > kernel: 524174 pages RAM > > The test sets up a swap on low-memory machines, however the detected > total memory on Standard_B1ms was 1.878 GB, so slightly below the 1.8 s/below/above/ Maybe this can be corrected when applying... Best regards, Krzysztof
Looks reasonable to me.
Acked-by: Po-Hsu Lin <po-hsu.lin@canonical.com>
Applied with s/below/above/ and pushed.
Thanks
Sam
diff --git a/ubuntu_lttng_smoke_test/ubuntu_lttng_smoke_test.py b/ubuntu_lttng_smoke_test/ubuntu_lttng_smoke_test.py index 434a5ac2e5b0..899d91652162 100755 --- a/ubuntu_lttng_smoke_test/ubuntu_lttng_smoke_test.py +++ b/ubuntu_lttng_smoke_test/ubuntu_lttng_smoke_test.py @@ -40,7 +40,7 @@ class ubuntu_lttng_smoke_test(test.test): df_gb = os.statvfs('/').f_bsize * os.statvfs('/').f_bavail / (1024.**3) cmd = 'swapon --show' swapon = utils.system_output(cmd) - if mem_gb < 1.8 and df_gb > 4.0 and swapon == '': + if mem_gb < 3.0 and df_gb > 4.0 and swapon == '': swap_file = '/tmp/swapfile' cmd = 'fallocate -l 2G %s' % swap_file utils.system(cmd)
The test was failing on Azure Standard_B1ms instance due to OOM killer: utils:0153| [stdout] gcc: fatal error: Killed signal terminated program cc1 In journalctl: kernel: cc1 invoked oom-killer: gfp_mask=0x100cca(GFP_HIGHUSER_MOVABLE), order=0, oom_score_adj=0 ... kernel: Free swap = 0kB kernel: Total swap = 0kB kernel: 524174 pages RAM The test sets up a swap on low-memory machines, however the detected total memory on Standard_B1ms was 1.878 GB, so slightly below the 1.8 threshold and the swap was not setup. Bump the threshold to really higher value, so this lack of memory won't hit us. Setting up swap (when not needed) should not impact the test, except making setup() slightly slower. BugLink: https://bugs.launchpad.net/bugs/1847989 Signed-off-by: Krzysztof Kozlowski <krzysztof.kozlowski@canonical.com> --- ubuntu_lttng_smoke_test/ubuntu_lttng_smoke_test.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)