Scalix 11 GA Bad Delete Request Preventing SIS Updates

[macdonae]

Scalix 11 GA, CentOS 4.4

I am having a serious problem with the SIS service indexing new messages. It seems to be stuck on a delete request with a dodgy dref. Here is some sample output from the access_log taken just after a omshut / omrc, showing where the indexer requests start to go awry:

/var/log/httpd/acces_log

192.168.1.30 - - [14/Jan/2007:10:50:04 +0000] "GET /sis/indexer?fn=add&uid=bruce.sis.thor.bkmacd.com&pdref=00013619de1d53db&dref=00011e845a6a8f4a&indexid=1e01a8c0-458284cc-459db3a3-192df&flags=seen,unflagged,unanswered,undeleted,undraft,unlabel1,unlabel2,unlabel3,unlabel4,unlabel5,unlabel6,unlabel7,unlabel8,unjunk,unnonjunk,unforwarded HTTP/1.1" 204 4 "-" "-"
192.168.1.30 - - [14/Jan/2007:10:50:04 +0000] "GET /sis/indexer?fn=add&uid=bruce.sis.thor.bkmacd.com&pdref=00013619de1d53db&dref=00011ebfbbe37873&indexid=1e01a8c0-458284cc-459e8947-19444&flags=seen,unflagged,unanswered,undeleted,undraft,unlabel1,unlabel2,unlabel3,unlabel4,unlabel5,unlabel6,unlabel7,unlabel8,unjunk,unnonjunk,unforwarded HTTP/1.1" 204 4 "-" "-"
192.168.1.30 - - [14/Jan/2007:10:50:04 +0000] "GET /sis/indexer?fn=add&uid=bruce.sis.thor.bkmacd.com&pdref=00013619de1d53db&dref=00011e73af314b91&indexid=1e01a8c0-458284cc-45a3691f-199f9&flags=seen,unflagged,unanswered,undeleted,undraft,unlabel1,unlabel2,unlabel3,unlabel4,unlabel5,unlabel6,unlabel7,unlabel8,unjunk,unnonjunk,unforwarded HTTP/1.1" 201 4 "-" "-"
192.168.1.30 - - [14/Jan/2007:10:50:04 +0000] "GET /sis/indexer?fn=delete&uid=bruce.sis.thor.bkmacd.com&dref=(\x85\xc9\x8b\x85h\xff\xff\xff\x0f\x95\xc2\x85\xc0\x0f\x95\xc0\t\xd0\xa8\x01\x0f\x85\xa0\x01&pdref=\t\xd0\xa8\x01\x0f\x85\xa0\x01 HTTP/1.1" 400 45 "-" "-"
192.168.1.30 - - [14/Jan/2007:10:50:14 +0000] "GET /sis/indexer?fn=delete&uid=bruce.sis.thor.bkmacd.com&dref=(\x85\xc9\x8b\x85h\xff\xff\xff\x0f\x95\xc2\x85\xc0\x0f\x95\xc0\t\xd0\xa8\x01\x0f\x85\xa0\x01&pdref=\t\xd0\xa8\x01\x0f\x85\xa0\x01 HTTP/1.1" 400 45 "-" "-"

... the same delete failure message is repeated every 10 seconds.

I then ran kill -SIGUSR2 on the indexer process as documented in the FAQ and it repeats the following ad-infinitum in /var/opt/scalix/tr/s/tmp/indexer.log:

[b]Logging Started
Awake liveRequests=100
WAIT_FOR_SIS DELETE user=103
dref=(?É??hÿÿÿ?Â?À?À Ã

[jch]

That's interesting. I wonder why it happens?

Anyway, to get around this:

shut down the scalix server
remove all the files in ~/indexwork (/var/opt/scalix/??/s/indexwork)
restart the server
run sxmkindex

jch

[jch]

In case you're interested, I've put a bug in for this: 14425.

We're not at all sure how you can get such an entry in the indexwork queue, let alone why the SIS thinks it's a temporary error.

jch

[macdonae]

That worked - thanks!

Bruce

[mitz]

FWIW, I had the same issue...which is interesting, because I purposely didn't install SIS.

Is there some process that I need to disable in that case?

mitz

Dell 2900 - 2G RAM
Fedora Core 5 x86_64
Scalix 11