Morten Stig Andersen wrote:
>> Has anyone got a solution to this:
>> The "PID" field of coding sequences in EMBL cannot be indexed in SRS.
> The srsupdate script can generate and compress the individual index
> files, but it dumps core when it tries to merge them. There are no
> problems in EMBLNEW, or with generation with any of the other EMBL
> indices. My SRS server runs on a SGI IRIX64, v. 6.4.
>> An extract from the log file is given below.
>> + srsbuild EMBL -f pid -xdir SRSINX: -odir SRSINX: -s unix -nn
> ...processing EMBL
> ...processing /usr/databases/embl/est1.dat
> ...processing /usr/databases/embl/est1.dat
>> ...wrote index file "/z/monet/appl/srs51/index/embl_pid_1.inx"
> size: 0 kbytes, bucket size: 1, records: 0
> ...wrote IDs file "/z/monet/appl/srs51/index/embl_pid_1.ids"
> size: 1 kbytes, IDs: 0
> ...processing /usr/databases/embl/est10.dat
> ...processing /usr/databases/embl/est10.dat
>> ...wrote index file "/z/monet/appl/srs51/index/embl_pid_2.inx"
> size: 1 kbytes, bucket size: 2, records: 3
>> (... output truncated here)
>> ...processing /usr/databases/embl/vrt.dat
> ...processing /usr/databases/embl/vrt.dat
>> ...wrote index file "/z/monet/appl/srs51/index/embl_pid_31.inx"
> size: 892 kbytes, bucket size: 40, records: 42034
> ...wrote IDs file "/z/monet/appl/srs51/index/embl_pid_31.ids"
> size: 421 kbytes, IDs: 42043
> ...program "srsbuild" completed successfully.
>> real 5h58m29.60s
> user 5h25m14.96s
> sys 17m58.02s
> + srsbuild EMBL -c -f pid -xdir SRSINX: -odir SRSINX: -s unix
> ...wrote IDs file "/z/monet/appl/srs51/index/embl_pid_1.ids"
> size: 1 kbytes, IDs: 0
> ...wrote IDs file "/z/monet/appl/srs51/index/embl_pid_2.ids"
> size: 1 kbytes, IDs: 3
>> (... output truncated here)
>> ...wrote IDs file "/z/monet/appl/srs51/index/embl_pid_31.ids"
> size: 253 kbytes, IDs: 42043
> ...program "srsbuild" completed successfully.
>> real 0m3.54s
> user 0m1.01s
> sys 0m0.87s
> + srsbuild EMBL -m -f pid -xdir SRSINX: -odir SRSINX: -s unix
> ./embl_pid[17]: 8624 Bus error(coredump)
>> real 1m0.01s
> user 0m33.25s
> sys 0m25.54s
>In my experience srs does not really like it if the first index in a "merge" situation is empty. Changing the order in which the different files are indexed,
(in embl.i), making sure that the first file has pids in its entries, might do the
trick. You will also have to rebuild all the indices again when you do that.
Cheers,
Martin
--
-------------------------------------------------------------------
| Martin Hilbers http://www.dci.clrc.ac.uk/Person.asp?m.p.hilbers |
| SEQNET | E-mail: m.p.hilbers at dl.ac.uk |
| Daresbury Laboratory | Tel: +44-1925-603492 |
| Daresbury, Warrington | Fax: +44-1925-603100 |
| Cheshire WA4 4AD | SEQNET is the UK national EMBNet node |
| United Kingdom | http://www.seqnet.dl.ac.uk/ |
-------------------------------------------------------------------
