Hi all,
thanks go to
Danny Cox
Daniel Baldoni
Rob Montjoy
we had no further errors so far and therefore did not undertake anything.
The machine seems to run fine despite the fact that it carries
a heavy load (oracle database in production environement).
So for now we leave it as it is.
BTW The mentioned patch 103981-16 was installed in a version -14 as
found on our latest sunsolve CD.
BTW2 We do not use VXFS.
Andreas
Here are the answers:
----------------------------
Man ..does THAT look famillar. Very interesting. One of my
customers was using E450/A5000 for their file servers, also
using Veritas VXFS and TAS. We had problems galore which
appear to be a problem with the PCI FC-AL card. This struggles
under load much as you describe here.
One thing you MUST do is make sure that the firmware revisions
of your disk controllers etc are up to date. It'll reduce this.
I can't guarantee it will fix it though! If you're running vxfs
then make sure that you have the recommended patches - there's
one patch which is slightly suspect and the recommendations are
that you use the previous version. I can probably get details if
you need more info there.
I'd be very interested to know how you go on here. It really did
cause me problems and hasn't been completely resolved yet.
Currently we've moved to an Ultra II with a SCSI multipack
instead!
----------------------------
I hate to tell you this but you're probably going to have to replace two of
your drives (sd30 and sd32). That is, if the machine needs to provide high
performance - but if it doesn't why buy an E450?.
Our new box exhibited the same symptoms (but under a fairly light I/O load)
and the official word from SUN was that the electronics on some drives just
couldn't keep up. They freely replaced our drive within a day (but we have
a maintenance contract).
----------------------------
This is most likely a hardware problem but you should install the latest
2.5.1 recommanded patches plus the latest glm patch (103981-16). If this
does not work call Sun and get them to replace the SCSI adaptor.
There was someone on the sun-managers list with a simliar problem with
a GLM scsi adaptor and they backed down the transfer rate is the .conf
but you should not have to do this.
---- And this was my original message:our new E450 (Solaris 2.5.1) showed under heavy disk activity the following entries in /var/adm/messages:
Dec 23 23:16:59 frondell unix: WARNING: /pci@1f,4000/scsi@4 (glm2): Dec 23 23:16:59 frondell unix: SCSI bus DATA IN phase parity error Dec 23 23:16:59 frondell unix: WARNING: ID[SUNWpd.glm.parity_check.6010] Dec 23 23:16:59 frondell unix: WARNING: /pci@1f,4000/scsi@4 (glm2): Dec 23 23:16:59 frondell unix: Target 0 reducing sync. transfer rate Dec 23 23:16:59 frondell unix: WARNING: ID[SUNWpd.glm.sync_wide_backoff.6014] Dec 23 23:16:59 frondell unix: WARNING: /pci@1f,4000/scsi@4/sd@0,0 (sd30): Dec 23 23:16:59 frondell unix: SCSI transport failed: reason 'tran_err': retrying command Dec 23 23:16:59 frondell unix:
And then again later:
Dec 23 23:19:57 frondell unix: WARNING: /pci@1f,4000/scsi@4 (glm2): Dec 23 23:19:57 frondell unix: SCSI bus DATA IN phase parity error Dec 23 23:19:57 frondell unix: WARNING: ID[SUNWpd.glm.parity_check.6010] Dec 23 23:19:57 frondell unix: WARNING: /pci@1f,4000/scsi@4 (glm2): Dec 23 23:19:57 frondell unix: Target 2 reducing sync. transfer rate Dec 23 23:19:57 frondell unix: WARNING: ID[SUNWpd.glm.sync_wide_backoff.6014] Dec 23 23:19:57 frondell unix: WARNING: /pci@1f,4000/scsi@4/sd@2,0 (sd32): Dec 23 23:19:57 frondell unix: SCSI transport failed: reason 'tran_err': retrying command Dec 23 23:19:57 frondell unix:
(followed by another series of such entries some minutes later). -- **************************************************************************** * Andreas Priebe E-Mail: Andreas.Priebe@promos-consult.de * * Promos consult GmbH & Co KG Tel/FAX: 030 243 227 -13 / -29 * ****************************************************************************
This archive was generated by hypermail 2.1.2 : Fri Sep 28 2001 - 23:13:14 CDT