[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[Xen-devel] Re: Timing out waiting for vbd device *only when xend started with trace_start*



Ewan Mellor wrote:
This is showing up intermittently with our regression tests.  It seems to be a
race of some sort, which would be consistent with it going away or coming back
when you fiddle with the logging -- all you are doing is changing the timing
of things so that it appears or does not appear.

If you export XENSTORED_TRACE=1 before starting Xend and Xenstored, then you
will get a log file in /var/log/xenstored-trace.log.  This information is
interesting to me, because it shows the ordering over time of the store
events, which makes it easier to piece together what's going on.

It does indeed feel like a race condition. Unfortunetly, setting XENSTORED_TRACE impacts the timing such as to make it nonreproducible.

Also, could you put a line in
/usr/lib/python/xen/xend/server/DevController.py, to log entry into the
hotplugStatus function (the syntax is log.debug("hotplugStatus entered"), and
it goes below the def hotplugStatus(): line).

Done; xend.log attached. No xenstored.trace yet for the reason described above.

Since the issue appears to be happening when xend is running below full speed but xenstored is not, I'm wondering if some mechanism to artificially slow xend (running it in valgrind?) would be effective in permitting the issue to be reproduced with tracing in place on both processes. I'll look into it, unless you have other suggestions.

Attachment: xend.log.bz2
Description: BZip2 compressed data

_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-devel

 


Rackspace

Lists.xenproject.org is hosted with RackSpace, monitoring our
servers 24x7x365 and backed by RackSpace's Fanatical Support®.