Autopktest broken on ppc64el (hirsute fio 3.21)

Bug #1903962 reported by Christian Ehrhardt  on 2020-11-12
6
This bug affects 1 person
Affects Status Importance Assigned to Milestone
fio (Ubuntu)
Undecided
Unassigned
multipath-tools (Ubuntu)
Undecided
Unassigned
tgt (Ubuntu)
Undecided
Unassigned

Bug Description

Example:
https://objectstorage.prodstack4-5.canonical.com/v1/AUTH_77e2ada1e7a84929a74ba3b87153c0ac/autopkgtest-hirsute/hirsute/ppc64el/m/multipath-tools/20201110_202353_71c1e@/log.gz

Error:
+ fio /tmp/autopkgtest.d8v3Mz/tgtbasedmpaths-artifacts/path-change-check.fio
verify-phase: (g=0): rw=read, bs=(R) 64.0KiB-64.0KiB, (W) 64.0KiB-64.0KiB, (T) 64.0KiB-64.0KiB, ioengine=sync, iodepth=1
fio-3.21
Starting 1 thread
Killed

As I outlined upstream
 https://github.com/axboe/fio/issues/1123
this is an OOM due to fio now consuming more memory.

Until there is a fix or upstream guidance available we will need to adapt our testing to use bigger systems to be able to run those successfully.

Related branches

Since it breaks the multipath-tools test I'll add a bug task just for tracking

tags: added: update-excuse
Changed in fio (Ubuntu):
status: New → Triaged

So far it "only" breaks the multipath-tools run on ppc64el, but I'm concerned that the other architectures are just surviving on a sliver of free memory (since I was able to reproduce the same fio behavior on all architectures).

Therefore I'll propose adding it to big_packages on all architectures, otherwise we too likely end up with a re-occurrence of this issue or a flaky test.

MP => https://code.launchpad.net/~paelzer/autopkgtest-cloud/+git/autopkgtest-cloud/+merge/393641

To post a comment you must log in.
This report contains Public information  Edit
Everyone can see this information.

Other bug subscribers