[Xen-devel] osstest lossage re linux-next 62999 and ms-queuedaemon

[Ian Jackson <Ian.Jackson@xxxxxxxxxxxxx>]

FYI:

I found the sg-report-flight for 62999 had been taking an inordinate
amount of time (>1h elapsed) (with big db locks held).  I have saved
its command line rune[1] for further investigation.

I also found that the ms-queuedaemon was spinning in recv().  I think
this was due to a bug in chan-read-data.  The bug is exposed only when
a client says it is going to send a certain number of bytes and then
fails to do so.  I don't see how this would arise in the production
colo.  Maybe a process ^Z'd halfway ?

I cowboyed a fix straight into daemons-testing.git and killed the
spinning instance so that it would restart.


Everything seems less wedged now.  I will investigate more properly on
Monday.

Ian.


[1]

  with-lock-ex -w /home/osstest/testing.git/report-lock ./sg-report-job-history 
--html-dir=/home/logs/results/ --flight=62999
+ ./sg-report-flight --html-dir=/home/logs/logs//62999/ --allow=allow.all 
--allow=allow.linux-next --blessings=real --info-headers 
--include-begin=tmp/62999.heading-info --machine-readable-output=linux-next.mro 
--this-linux=cd685d8558e92f3d3ba7e070ac03ae2585f70ba1 
--that-linux=5b5f1455272e23f4e7889cec37228802d8d01adf 
--branches-also=linux-linus 62999

which when turned into a test case probably looks like this

  ./sg-report-flight --debug --allow=allow.all --allow=allow.linux-next 
--blessings=real --info-headers 
--this-linux=cd685d8558e92f3d3ba7e070ac03ae2585f70ba1 
--that-linux=5b5f1455272e23f4e7889cec37228802d8d01adf 
--branches-also=linux-linus 62999

_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxx
http://lists.xen.org/xen-devel