On Wed, Mar 24, 2021 at 1:24 PM Alan G <alan+ovirt@griff.me.uk> wrote:
Looking back in the logs, in fact the first error we get is Out of memory. So it seems we're hitting https://bugzilla.redhat.com/show_bug.cgi?id=1623851

It's not clear from the ticket. Is there an explicit fix for this is 4.4, or the problem just kind of went away?

If it is the described issue, the problem seems to go away in 4.4. The reason might be a newer kernel and libnl3.


 



---- On Wed, 24 Mar 2021 11:18:57 +0000 Alan G <alan+ovirt@griff.me.uk> wrote ----

Hi,

I sent this a while back and never got a response. We've since upgrade to 4.3 and the issue persists.

2021-03-24 10:53:48,934+0000 ERROR (periodic/2) [virt.periodic.Operation] <vdsm.virt.sampling.HostMonitor object at 0x7f5964398350> operation failed (periodic:188)
Traceback (most recent call last):
  File "/usr/lib/python2.7/site-packages/vdsm/virt/periodic.py", line 186, in __call__
    self._func()
  File "/usr/lib/python2.7/site-packages/vdsm/virt/sampling.py", line 481, in __call__
    stats = hostapi.get_stats(self._cif, self._samples.stats())
  File "/usr/lib/python2.7/site-packages/vdsm/host/api.py", line 50, in get_stats
    decStats = stats.produce(first_sample, last_sample)
  File "/usr/lib/python2.7/site-packages/vdsm/host/stats.py", line 72, in produce
    stats.update(get_interfaces_stats())
  File "/usr/lib/python2.7/site-packages/vdsm/host/stats.py", line 154, in get_interfaces_stats
    return net_api.network_stats()
  File "/usr/lib/python2.7/site-packages/vdsm/network/api.py", line 63, in network_stats
    return netstats.report()
  File "/usr/lib/python2.7/site-packages/vdsm/network/netstats.py", line 32, in report
    stats = link_stats.report()
  File "/usr/lib/python2.7/site-packages/vdsm/network/link/stats.py", line 34, in report
    for iface_properties in iface.list():
  File "/usr/lib/python2.7/site-packages/vdsm/network/link/iface.py", line 257, in list
    for properties in itertools.chain(link.iter_links(), dpdk_links):
  File "/usr/lib/python2.7/site-packages/vdsm/network/netlink/link.py", line 47, in iter_links
    with _nl_link_cache(sock) as cache:
  File "/usr/lib64/python2.7/contextlib.py", line 17, in __enter__
    return self.gen.next()
  File "/usr/lib/python2.7/site-packages/vdsm/network/netlink/__init__.py", line 108, in _cache_manager
    cache = cache_allocator(sock)
  File "/usr/lib/python2.7/site-packages/vdsm/network/netlink/link.py", line 157, in _rtnl_link_alloc_cache
    return libnl.rtnl_link_alloc_cache(socket, AF_UNSPEC)
  File "/usr/lib/python2.7/site-packages/vdsm/network/netlink/libnl.py", line 578, in rtnl_link_alloc_cache
    raise IOError(-err, nl_geterror(err))
IOError: [Errno 16] Message sequence number mismatch

This occurs on both nodes in the cluster. A restart of vdsm/supervdsm will sort it for a while, but within 24 hours it occurs again. We run a number of clusters and it only occurs on one so must be some specific corner case we're triggering.

I can find almost no information on this. The best I could find was this https://linuxlizard.com/2020/10/18/message-sequence-number-mismatch-in-libnl/ which details a sequence number issue. I'm guessing I'm experiencing the same issue in that the nl sequence numbers are getting out of sync and closing/re-opening the nl socket (aka restart vdsm) is the only way to resolve.

I've completely hit a brick wall with it. We've had to disable fencing on both nodes as sometimes they get erroneously fenced when vdsm stops function correctly. At this point I'm thinking about replaced the severs with different models in-case it's something in the NIC drivers...

Alan


---- On Mon, 06 Jan 2020 10:54:52 +0000 Alan G <alan+ovirt@griff.me.uk> wrote ----

Hi,

I have issues with one host where supervdsm is failing in network_caps.

I see the following trace in the log.

MainProcess|jsonrpc/1::ERROR::2020-01-06 03:01:05,558::supervdsm_server::100::SuperVdsm.ServerCallback::(wrapper) Error in network_caps
Traceback (most recent call last):
  File "/usr/lib/python2.7/site-packages/vdsm/supervdsm_server.py", line 98, in wrapper
    res = func(*args, **kwargs)
  File "/usr/lib/python2.7/site-packages/vdsm/network/api.py", line 56, in network_caps
    return netswitch.configurator.netcaps(compatibility=30600)
  File "/usr/lib/python2.7/site-packages/vdsm/network/netswitch/configurator.py", line 317, in netcaps
    net_caps = netinfo(compatibility=compatibility)
  File "/usr/lib/python2.7/site-packages/vdsm/network/netswitch/configurator.py", line 325, in netinfo
    _netinfo = netinfo_get(vdsmnets, compatibility)
  File "/usr/lib/python2.7/site-packages/vdsm/network/netinfo/cache.py", line 150, in get
    return _stringify_mtus(_get(vdsmnets))
  File "/usr/lib/python2.7/site-packages/vdsm/network/netinfo/cache.py", line 59, in _get
    ipaddrs = getIpAddrs()
  File "/usr/lib/python2.7/site-packages/vdsm/network/netinfo/addresses.py", line 72, in getIpAddrs
    for addr in nl_addr.iter_addrs():
  File "/usr/lib/python2.7/site-packages/vdsm/network/netlink/addr.py", line 33, in iter_addrs
    with _nl_addr_cache(sock) as addr_cache:
  File "/usr/lib64/python2.7/contextlib.py", line 17, in __enter__
    return self.gen.next()
  File "/usr/lib/python2.7/site-packages/vdsm/network/netlink/__init__.py", line 92, in _cache_manager
    cache = cache_allocator(sock)
  File "/usr/lib/python2.7/site-packages/vdsm/network/netlink/libnl.py", line 469, in rtnl_addr_alloc_cache
    raise IOError(-err, nl_geterror(err))
IOError: [Errno 16] Message sequence number mismatch

A restart of supervdsm will resolve the issue for a period, maybe 24 hours, then it will occur again. So I'm thinking it's resource exhaustion or a leak of some kind?

Running 4.2.8.2 with VDSM at 4.20.46.

I've had a look through the bugzilla and can't find an exact match, closest was this one https://bugzilla.redhat.com/show_bug.cgi?id=1666123 which seems to be a RHV only fix.

Thanks,

Alan

_______________________________________________
Users mailing list -- users@ovirt.org
To unsubscribe send an email to users-leave@ovirt.org


_______________________________________________
Users mailing list -- users@ovirt.org
To unsubscribe send an email to users-leave@ovirt.org


_______________________________________________
Users mailing list -- users@ovirt.org
To unsubscribe send an email to users-leave@ovirt.org
Privacy Statement: https://www.ovirt.org/privacy-policy.html
oVirt Code of Conduct: https://www.ovirt.org/community/about/community-guidelines/
List Archives: https://lists.ovirt.org/archives/list/users@ovirt.org/message/4525CHJL7E7AXXB6Q5VG4Q7LRFFB2ILL/


--

Ales Musil

Software Engineer - RHV Network

Red Hat EMEA

amusil@redhat.com    IM: amusil