Hi, I have 2 IBM x346 eServers, both of them with 2 Qlogic HBAs (qla2340). They are connected rendundantly to a IBM DS4300 (fiber storage). I have Oracle 10gR2 RAC installed. RHEL4 U2 x86_64. My problem is the following: Each server works very well, if alone,ie, if the other one has not mounted the ocfs2 partition. But when both of them are working, the performance goes to hell. For example: there is a job that, when it runs on one server alone, takes 4 seconds, but when I put both servers on line, it takes 3 minutes. What could it be ? Is there any conf that I could check. I was thinking about change the ocfs2 for other fs (gfs or other)... could it be the case ? Thanks in advance. Marcio Costa. -- Open WebMail Project (http://openwebmail.org)
On Wed, 2006-02-08 at 12:51 -0300, Marcio Costa wrote:> Hi, > > I have 2 IBM x346 eServers, both of them with 2 Qlogic HBAs (qla2340). They > are connected rendundantly to a IBM DS4300 (fiber storage). I have Oracle > 10gR2 RAC installed. RHEL4 U2 x86_64. My problem is the following: > > Each server works very well, if alone,ie, if the other one has not mounted > the ocfs2 partition. But when both of them are working, the performance goes > to hell. > For example: there is a job that, when it runs on one server alone, takes 4 > seconds, but when I put both servers on line, it takes 3 minutes. > > What could it be ? Is there any conf that I could check. I was thinking > about change the ocfs2 for other fs (gfs or other)... could it be the case ? > > Thanks in advance.I just got this email today but it was dated Feb 8th. Were you able to resolve this issue? I'm going down the same road in the next few months and I'd like to know about any 10gR2 x86_64 issues with OCFS2. Thanks. /Brian/ -- Brian Long | | | IT Data Center Systems | .|||. .|||. Cisco Linux Developer | ..:|||||||:...:|||||||:.. Phone: (919) 392-7363 | C i s c o S y s t e m s
Marcio Costa wrote:>Hi, > >I have 2 IBM x346 eServers, both of them with 2 Qlogic HBAs (qla2340). They >are connected rendundantly to a IBM DS4300 (fiber storage). I have Oracle >10gR2 RAC installed. RHEL4 U2 x86_64. My problem is the following: > >Each server works very well, if alone,ie, if the other one has not mounted >the ocfs2 partition. But when both of them are working, the performance goes >to hell. >For example: there is a job that, when it runs on one server alone, takes 4 >seconds, but when I put both servers on line, it takes 3 minutes. > > >For starters, I would suggest poking into the system with the em dbconsole. Run ADDM for the time period in question and see what it complains about. That should at least give you some idea where the problem is.>What could it be ? Is there any conf that I could check. I was thinking >about change the ocfs2 for other fs (gfs or other)... could it be the case ? > >Thanks in advance. > >Marcio Costa. > >-- >Open WebMail Project (http://openwebmail.org) > >_______________________________________________ >Ocfs2-users mailing list >Ocfs2-users at oss.oracle.com >http://oss.oracle.com/mailman/listinfo/ocfs2-users > > > >
Jizhong_Wang@whirlpool.com
2006-Mar-22 15:20 UTC
[Ocfs2-users] RHEL4 x86_64 OCFS2 10gR2 performance
I had a similar situation with similar hardware settings (IBM eSeries, SHARK, Gigabit NICs for interconnect). What I found out was I picked both public and private NICs in VIP configuration (GUI) and most inter-instance chattings were using the public interfaces. After having corrected the problem (I re-installed RAC), I got much-better-performaning systems Jizhong Wang Kevin Hulse <kevin.hulse at placemark.com> Sent by: ocfs2-users-bounces at oss.oracle.com 03/22/2006 09:34 AM To Marcio Costa <marciopc at interativainternet.com.br> cc "ocfs2-users at oss.oracle.com" <ocfs2-users at oss.oracle.com>, josue at interativainternet.com.br Subject Re: [Ocfs2-users] RHEL4 x86_64 OCFS2 10gR2 performance Marcio Costa wrote:>Hi, > >I have 2 IBM x346 eServers, both of them with 2 Qlogic HBAs (qla2340).They>are connected rendundantly to a IBM DS4300 (fiber storage). I have Oracle>10gR2 RAC installed. RHEL4 U2 x86_64. My problem is the following: > >Each server works very well, if alone,ie, if the other one has notmounted>the ocfs2 partition. But when both of them are working, the performancegoes>to hell. >For example: there is a job that, when it runs on one server alone, takes4>seconds, but when I put both servers on line, it takes 3 minutes. > > >For starters, I would suggest poking into the system with the em dbconsole. Run ADDM for the time period in question and see what it complains about. That should at least give you some idea where the problem is.>What could it be ? Is there any conf that I could check. I was thinking >about change the ocfs2 for other fs (gfs or other)... could it be thecase ?> >Thanks in advance. > >Marcio Costa. > >-- >Open WebMail Project (http://openwebmail.org) > >_______________________________________________ >Ocfs2-users mailing list >Ocfs2-users at oss.oracle.com >http://oss.oracle.com/mailman/listinfo/ocfs2-users > > > >_______________________________________________ Ocfs2-users mailing list Ocfs2-users at oss.oracle.com http://oss.oracle.com/mailman/listinfo/ocfs2-users -------------- next part -------------- An HTML attachment was scrubbed... URL: http://oss.oracle.com/pipermail/ocfs2-users/attachments/20060322/5d7a2470/attachment.html
Reinstalling CRS or the RDBMS to repair this sort of
issue is dramatic overkill. You can reconfigure the NICs
with the oifcfg utility pretty easily. I had a misconfigured
cluster interconnect and fixed it that way without
reinstalling anything.
This problem had manifested itself as poor cluster
transfer response times which I detected through the
dbconsole. I discovered the root of the problem by
running sar and examining the network statistics.
Jizhong_Wang at whirlpool.com wrote:
>
> I had a similar situation with similar hardware settings (IBM eSeries,
> SHARK, Gigabit NICs for interconnect). What I found out was I picked
> both public and private NICs in VIP configuration (GUI) and most
> inter-instance chattings were using the public interfaces. After
> having corrected the problem (I re-installed RAC), I got
> much-better-performaning systems
>
>
> Jizhong Wang
>
>
>
>
> *Kevin Hulse <kevin.hulse at placemark.com>*
> Sent by: ocfs2-users-bounces at oss.oracle.com
>
> 03/22/2006 09:34 AM
>
>
> To
> Marcio Costa <marciopc at interativainternet.com.br>
> cc
> "ocfs2-users at oss.oracle.com" <ocfs2-users at
oss.oracle.com>,
> josue at interativainternet.com.br
> Subject
> Re: [Ocfs2-users] RHEL4 x86_64 OCFS2 10gR2 performance
>
>
>
>
>
>
>
>
>
> Marcio Costa wrote:
>
> >Hi,
> >
> >I have 2 IBM x346 eServers, both of them with 2 Qlogic HBAs
> (qla2340). They
> >are connected rendundantly to a IBM DS4300 (fiber storage). I have
> Oracle
> >10gR2 RAC installed. RHEL4 U2 x86_64. My problem is the following:
> >
> >Each server works very well, if alone,ie, if the other one has not
> mounted
> >the ocfs2 partition. But when both of them are working, the
> performance goes
> >to hell.
> >For example: there is a job that, when it runs on one server alone,
> takes 4
> >seconds, but when I put both servers on line, it takes 3 minutes.
> >
> >
> >
>
> For starters, I would suggest poking into the system with
> the em dbconsole. Run ADDM for the time period in question
> and see what it complains about. That should at least give
> you some idea where the problem is.
>
> >What could it be ? Is there any conf that I could check. I was thinking
> >about change the ocfs2 for other fs (gfs or other)... could it be the
> case ?
> >
> >Thanks in advance.
> >
> >Marcio Costa.
> >
> >--
> >Open WebMail Project (http://openwebmail.org)
> >
> >_______________________________________________
> >Ocfs2-users mailing list
> >Ocfs2-users at oss.oracle.com
> >http://oss.oracle.com/mailman/listinfo/ocfs2-users
> >
> >
> >
> >
>
>
> _______________________________________________
> Ocfs2-users mailing list
> Ocfs2-users at oss.oracle.com
> http://oss.oracle.com/mailman/listinfo/ocfs2-users
>