Hi Gareth, Winnie
Thanks for your help. Same issue has been reported from other NGI's so it is not an UK only problem. It seems that some topbdii's are caching information for quite long time. I checked all three UK top bdii's and Manchester top bdii is still publishing mq.afroditi.hellasgrid.gr
ldapsearch -x -LLL -h top-bdii.tier2.hep.manchester.ac.uk -p 2170 -b o=grid '(&(objectClass=GlueService)(GlueServiceType=msg.broker.stomp))' msg.broker.stomp
dn: GlueServiceUniqueID=mq.cro-ngi.hr_msg.broker.stomp_3523291347,Mds-Vo-name=
egee.srce.hr,Mds-Vo-name=local,o=grid
dn: GlueServiceUniqueID=msg-test.cro-ngi.hr_msg.broker.stomp_3276635265,Mds-Vo
-name=egee.srce.hr,Mds-Vo-name=local,o=grid
dn: GlueServiceUniqueID=mq.afroditi.hellasgrid.gr_msg.broker.stomp_4075209668,
Mds-Vo-name=HG-03-AUTH,Mds-Vo-name=local,o=grid
I think, now we understand main issue and probably it will go away when all top bdii's expire its cache.
Cheers
Kashif
> -----Original Message-----
> From: Testbed Support for GridPP member institutes [mailto:TB-
> [log in to unmask]] On Behalf Of Winnie Lacesso
> Sent: 14 July 2014 10:49
> To: [log in to unmask]
> Subject: Re: "Upstream" problem for UKI?
>
> Dr Kashif wrote:
> > If any site has an ops jobs running for long time, can you please run
> > netstat on the WN where job is running to see whether it is hanging on
> > some connection? One of the Greek sites hosting message broker had a
> > power cut on Saturday
>
> Well I didn't run netstat, but found the process tree end at python hung on
> contacting host
>
> connect(3, {sa_family=AF_INET, sin_port=htons(6163),
> sin_addr=inet_addr("195.251.55.91")}, 16
>
> which is mq.afroditi.hellasgrid.gr
>
> Could you be persuaded to answer what does No handlers could be found
> for logger "stomp.py"
> (in gridjob.out) mean?
>
> Winnie Lacesso / Bristol University Particle Physics Computing Systems HH
> Wills Physics Laboratory, Tyndall Avenue, Bristol, BS8 1TL, UK
|