arm64 clustering hitting timeout (ADT failure)

[Kleber Sacilotto de Souza]

Testing failed on:
    arm64: https://objectstorage.prodstack4-5.canonical.com/v1/AUTH_77e2ada1e7a84929a74ba3b87153c0ac/autopkgtest-bionic/bionic/arm64/l/lxd/20191010_151713_6f6f4@/log.gz

[Thadeu Lima de Souza Cascardo (cascardo)]

This has been failing a lot with all of our kernels, and even when util-linux or shadow is the trigger.

It seems to fail on the clustering membership test, at least on two different instance failures I checked.

Compared to amd64, I see some nodes are offline on arm64 with "no heartbeat".

+ /usr/bin/lxc cluster list --verbose
+-------+-------------------------+----------+---------+----------------------------------+
| NAME | URL | DATABASE | STATE | MESSAGE |
+-------+-------------------------+----------+---------+----------------------------------+
| node1 | https://10.1.1.101:8443 | YES | ONLINE | fully operational |
+-------+-------------------------+----------+---------+----------------------------------+
| node2 | https://10.1.1.102:8443 | YES | OFFLINE | no heartbeat since 51.32886019s |
+-------+-------------------------+----------+---------+----------------------------------+
| node3 | https://10.1.1.103:8443 | YES | OFFLINE | no heartbeat since 54.868743549s |
+-------+-------------------------+----------+---------+----------------------------------+
| node4 | https://10.1.1.104:8443 | NO | ONLINE | fully operational |
+-------+-------------------------+----------+---------+----------------------------------+
| node5 | https://10.1.1.105:8443 | NO | ONLINE | fully operational |
+-------+-------------------------+----------+---------+----------------------------------+

[Stéphane Graber (stgraber)]

3.0.5 should be a bit more stable in that regard, when we get around to pushing it.
Until then, yes, armhf/arm64 will tend to be a bit more racy due to slow crypto on some CPUs.

[Launchpad Janitor (janitor)]

Status changed to 'Confirmed' because the bug affects multiple users.