605 error
Mahima Warrier
>605's are normally caused by hardware failure - its rare for 605's
>to be caused by the latest server software (nowdays) thru the fault
>of sybase (but beware 10.0.x on Solaris 2.4)
>There have in the distant past been occurances of 605's caused by
>other reasons - what's your version?
>Its been suggested that 605's could be caused by "dump tran with no_log"
>but i don't think i've ever seen one (in 5yrs).
>Hardware failure or baddy defined partitons/unix file systems are
>the normal cause.
>
>---
> ---------------------------------------------------------------
> | David J Ferrington The art of RDBMS's is a "Syance"! |
> | Swiss Bank Corp |
> | London Email: fer...@gb.swissbank.com |
> |--------------------------------------------------------------
Our configuration is Sybase 10.0.1 on SS-10 (Sun OS 4.1.3_U1)
We have database and log on seperate partitions (all raw) and use asynchronous
i/o. This machine is a dedicated dataserver. We get 605 errors at least once
a month. But our OS does not report any irregularity on the days that we get
the 605 error. In fact our OS has not reported any irregularity so far .
How then can we infer that it is caused due to Hardware failure ?
------------------------------------------------------------------------------
Mahima Warrier Work: 91-22-4300531
Centre for Monitoring Indian Economy, Bombay Fax: 91-22-4370558
mah...@cmie.ernet.in Home: 91-22-6295367
--
------------------------------------------------------------------------------
Mahima Warrier Work: 91-22-4300531
Centre for Monitoring Indian Economy, Bombay Fax: 91-22-4370558
mah...@cmie.ernet.in Home: 91-22-6295367
Wiz Pop
: David Ferrington writes:
: >605's are normally caused by hardware failure - its rare for 605's
: >to be caused by the latest server software (nowdays) thru the fault
: >of sybase (but beware 10.0.x on Solaris 2.4)
: Our configuration is Sybase 10.0.1 on SS-10 (Sun OS 4.1.3_U1)
: We have database and log on seperate partitions (all raw) and use
asynchronous
: i/o. This machine is a dedicated dataserver. We get 605 errors at least once
: a month. But our OS does not report any irregularity on the days that we get
: the 605 error. In fact our OS has not reported any irregularity so far .
: How then can we infer that it is caused due to Hardware failure ?
Are there 'real' 605's after the report? ie, does a dbcc checktable show
any errors on the table in question? If not, then you probably have a bug
in the handling of asynchronous i/o. I found mine after going to sybase
10.0.1 on the SGI platform. I bypassed it by turning off asynch i/o (use
switch -T1604 (I think) with the startserver. (I'll have to check that
and get back to you). I have been told that the bug with asynch i/o is
fixed in 10.0.2. Those of of not on the primary port list (sigh) have to
wait a while for that!
--
Don (hami...@uoguelph.ca), Librarian by Osmosis
McLaughlin Library, University of Guelph, Guelph, Ontario
Let P=practice, T=theory. Then x in T implies T = P. But x in P not T = P.
Frank Tiemann
into the same database simultaneously.
Frank
c...@sybase.com
In article <D4C6o...@cmie.ernet.in>, <mah...@cmie.ernet.in> writes:
Stuff deleted...
> Our configuration is Sybase 10.0.1 on SS-10 (Sun OS 4.1.3_U1)
> We have database and log on seperate partitions (all raw) and use
asynchronous
> i/o. This machine is a dedicated dataserver. We get 605 errors at least once
> a month. But our OS does not report any irregularity on the days that we get
> the 605 error. In fact our OS has not reported any irregularity so far .
> How then can we infer that it is caused due to Hardware failure ?
Make sure that cylinder 0 is not being used by a disk partition that Sybase
uses. Cylinder 0 contains the partition map for the disk. After a reboot, the
device (disk) doesn't exist to the operating system, because the partition map
is overwritten by Sybase...or any database for that fact. I'm not aware of any
databases that specifically check for cylinder 0.
Just a stab in the dark.
Erland Sommarskog
>Also, there is a bug creating 605's if you run multiple BCPs
>into the same database simultaneously.
If this is the thread "how to get a 605 error", I can report another
that hit me the other day. I tried to apply EBF 3194(?) but it filled
up my log in master. So I ran buildmaster and then LOAD DATABASE where-
upon I started my server which died immediately with 605.
Reading the trouble-shooting guide helped me out. This is an AIX box
on which Sybase for some very mysterious reason thinks that CP850 is
a suitable character set (CP850 is highly unsuitable whereever it
appears as it is non-standard). Naturally we had mended Sybase errone-
ous ways by chaning the character set and sort order to iso_1, and
furthermore deleted cp850 from syscharsets. (Else the server will
start to convert the deluded clients who also believes CP850 is cool.)
Anyway, apparently there is a lot of mumbo-jumbo with syscharsets -
in fact the EBF in question apparently fixes a bug with hard-coded
references to syscharsets, making it impossible to move a master
device. So if you have changed your sort order or character set or
mingled with syscharsets, you need to redo that before you load
back your dump of master.
--
Erland Sommarskog, som...@enea.se, Stockholm
John Anest
Another 605 generator:
SQLServer 10.01 for Netware (10.02 still ain't out) generates a 605
error any time you try to load the database from a dump file. Nice heh?
Note that there's no other way to send out upgrades to client systems
without revealing source.( you update syscomments text = null before
you dump the db initially ). Nice heh?
--
John Anest