| Author |
Message
|
| cactus |
 Posted: Wed Sep 28, 2005 8:18 am Post subject: SDR CHL goes in retry mode |
 |
|
Apprentice
Joined: 12 Oct 2002
Posts: 30
|
Hi,
We have 2 win 2003 servers A & B with same image and got same QMgr name .They are 2 separate servers one is production and other one as a standby.2 QMgrs running all the time and connects to same gateway on same port.Gateway SDR chl to connects to A(Prod) server and runs fine.Whenever SDR chl on B(Standby) starts to gateway(goes to retry due to mismatch of seqence number,we know) resulting SDR chl on gateway to A(Prod) goes into retry ..saying
09/27/05 22:18:24
AMQ9206: Error sending data to host "A production".
EXPLANATION:
An error occurred sending data over TCP/IP to "A server" . This may
be due to a communications failure.
ACTION:
The return code from the TCP/IP(write) call was 32 X('20'). Record these values
and tell your systems administrator.
----- amqccita.c : 2049 -------------------------------------------------------
09/27/05 22:18:24
AMQ9999: Channel program ended abnormally.
EXPLANATION:
Channel program 'GATEWAY TO A ' ended abnormally.
ACTION:
Look at previous error messages for channel program 'GATEWAY TO A' in theerror files to determine the cause of the failure.
When I stop the SDR chl from B to gateway.SDR chl gateway to A recovers automatically and runs fine.This is happening after reinstallation of MQ on B (standby server) earlier it was fine.verified A & B everthing OS and QMgr paremeters are same.
Server name and chl names were edited for easy understanding.
your inputs are highly appreciated.
thankx
cactus |
|
| Back to top |
|
 |
| hopsala |
| Posted: Wed Sep 28, 2005 9:04 am Post subject: |
|
|
Guardian
Joined: 24 Sep 2004
Posts: 960
|
Let me get this straight: channel GATE.TO.A runs fine, and when you start channel B.TO.GATE suddenly GATE.TO.A goes haywire, right? (it's a little hard to follow the way you wrote it...)
That does sound odd, but I have a theory - are you possibly using the same ip address for A and B? If so, then of course you have problems, god knows what routers do when they have two destinations for the same ip; my guess - some packets sent here, some there, and naturally there's trouble.
Second theory - are you using clusters?
Any FDCs hanging about on A,B or GATEWAY ? |
|
| Back to top |
|
|
| hopsala |
| Posted: Wed Sep 28, 2005 9:10 am Post subject: Re: SDR CHL goes in retry mode |
|
|
Guardian
Joined: 24 Sep 2004
Posts: 960
|
| cactus wrote: |
| We have 2 win 2003 servers A & B with same image and got same QMgr name .They are 2 separate servers one is production and other one as a standby. |
Btw, you do know this is a truely *terrible* way to implement workload balacing and failover? I mean, there are worse ways, but not many. I advise you to change this as soon as you can to a proper workload setup.
Search around the site about info on how to do this, or read the manuals concerning MSCS clusters with WMQ clusters. |
|
| Back to top |
|
|
| kevinf2349 |
| Posted: Wed Sep 28, 2005 9:13 am Post subject: |
|
|
Grand Master
Joined: 28 Feb 2003
Posts: 1311
Location: USA
|
| Quote: |
| We have 2 win 2003 servers A & B with same image and got same QMgr name |
Are you trying to talk to 2 queue managers of the same name on different boxes?
If so I believe the answer is..."no can do."...certainly not at the same time at least  |
|
| Back to top |
|
|
| hopsala |
| Posted: Wed Sep 28, 2005 9:23 am Post subject: |
|
|
Guardian
Joined: 24 Sep 2004
Posts: 960
|
| kevinf2349 wrote: |
Are you trying to talk to 2 queue managers of the same name on different boxes?
If so I believe the answer is..."no can do."...certainly not at the same time at least |
Are you sure?
I mean, of course he'll have a terrible time defining QRs and such, and of course this is not the way to go about doing anything, but I see no reason why it shouldn't work. I don't think that mq "knows" anything about the qms his channels connect to; MCA work entirely seperately from the kernel. |
|
| Back to top |
|
|
| cactus |
| Posted: Wed Sep 28, 2005 10:58 am Post subject: |
|
|
Apprentice
Joined: 12 Oct 2002
Posts: 30
|
| Quote: |
| channel GATE.TO.A runs fine, and when you start channel B.TO.GATE suddenly GATE.TO.A goes haywire, right? |
yes,that's right..!(I'm interested to know why its happening now..It was not happening before)
| Quote: |
| are you possibly using the same ip address for A and B? |
No,A & B got 2 diff ip's.
| Quote: |
| Second theory - are you using clusters? |
No,its point to point and its not in microsoft hardware cluster .
| Quote: |
| Any FDCs hanging about on A,B or GATEWAY ? |
no FDC's logged
| Quote: |
| you do know this is a truely *terrible* way to implement workload balacing and failover? |
We are not doing workload balancing or failover.B server is just a stand by.
| Quote: |
| Are you trying to talk to 2 queue managers of the same name on different boxes? |
yes,At a time gateway talks to only one Qmgr .If we are doing swicthing of the server from A to B.We change the conname parameter on SDR chl of gateway accordingly. |
|
| Back to top |
|
|
| Anirud |
| Posted: Wed Sep 28, 2005 11:34 am Post subject: Re: SDR CHL goes in retry mode |
|
|
Master
Joined: 12 Feb 2004
Posts: 285
Location: Vermont
|
| cactus wrote: |
| Whenever SDR chl on B(Standby) starts to gateway(goes to retry due to mismatch of seqence number,we know) resulting SDR chl on gateway to A(Prod) goes into retry |
Why are you starting the SDR channel on B(Standby) when the channel from A is already running?
Are the names of SDR channels on A and B the same?
If yes, I think when you start the channel on B, the corresponding RCVR channel on your gateway starts (which will be the same RCVR channel for A) and hence channel on A is retrying and this can be expected to happen.
You can check the IP address on the RCVR channel. It will let you know which server (A or B) the RCVR channel is talking to.
Hope this helps. |
|
| Back to top |
|
|
| cactus |
| Posted: Wed Sep 28, 2005 11:57 am Post subject: |
|
|
Apprentice
Joined: 12 Oct 2002
Posts: 30
|
| Quote: |
Why are you starting the SDR channel on B(Standby) when the channel from A is already running?
|
everyday mq backup batch scripts starts the chls to clear any msgs in xmitq(there won't be any) and stops it and backup is happening for few years.
| Quote: |
Are the names of SDR channels on A and B the same?
|
yes ,It got same image.
A Server - CLONE QMGR - Prod
B Server - CLONE QMGR - Stand-by
On B(Stand-By) server SDR CHL CLONEtoGATEWAY going to retry status - expected becoz mismatch in seq number and requester chl on gateway is connected to SDR chl on A(Prod) server .
On gateway SDR CHL GATEWAYtoCLONE connecting to A(Prod) server going to retry status - expected ?
If yes ..can u pls explain and moreover it was not happening before.It all started when reinstallation of MQ 5.3 csd10 done on B(stand-by) server.
Last edited by cactus on Wed Sep 28, 2005 12:05 pm; edited 1 time in total |
|
| Back to top |
|
|
| hopsala |
| Posted: Wed Sep 28, 2005 1:50 pm Post subject: |
|
|
Guardian
Joined: 24 Sep 2004
Posts: 960
|
| cactus wrote: |
| We are not doing workload balancing or failover.B server is just a stand by. |
Stand by is an a way to implement failover, thus you are implementing failover; if you wish you can call it "backup". Regardless of what it's called, this is a highly improper use of MQ, expect many other problems in the years to come, some immediate, others more subtle...
I wonder, what is the buisness need that drove you to such a configuration? |
|
| Back to top |
|
|
| Anirud |
| Posted: Wed Sep 28, 2005 1:51 pm Post subject: |
|
|
Master
Joined: 12 Feb 2004
Posts: 285
Location: Vermont
|
my apologies. I misread your post 
I thought the SDR channel from Server A to Gateway was retrying. Hence my comment ( that this can be expected).
Not sure why your channel from Gateway to CLONE (server A - Prod) is retrying. |
|
| Back to top |
|
|
| fjb_saper |
| Posted: Wed Sep 28, 2005 7:45 pm Post subject: |
|
|
Grand High Poobah
Joined: 18 Nov 2003
Posts: 20756
Location: LI,NY
|
If you want to fix it you need different channels:
Of course then you need to specify the channel for the xmitq because you'll have the same xmitq plugged into 2 different qmgrs/channels.
You can then switch the channel by altering the channel name on the xmitq.
Enjoy  |
|
| Back to top |
|
|
| kevinf2349 |
| Posted: Thu Sep 29, 2005 5:16 am Post subject: |
|
|
Grand Master
Joined: 28 Feb 2003
Posts: 1311
Location: USA
|
| Quote: |
kevinf2349 wrote:
Are you trying to talk to 2 queue managers of the same name on different boxes?
If so I believe the answer is..."no can do."...certainly not at the same time at least
Are you sure?
|
Hopsla
Well I guess it really depends on how reliable you want things. Lets take the case where you have transmission queues that are the same name as the reply to queue manager name......which one gets the reply and how do you know for sure if you share transmission queues across channel definitions?
We have a case where we have two indentically named queue managers (a stupid vendor restriction because they don't understand MQ) and we can 'talk to' both....but not at the same time.
I believe IBM's suggestion is not to have two queue managers of the same name in the same 'MQ network'.
I could be wrong, usually am actually  |
|
| Back to top |
|
|
| hopsala |
| Posted: Thu Sep 29, 2005 6:08 am Post subject: |
|
|
Guardian
Joined: 24 Sep 2004
Posts: 960
|
| kevinf2349 wrote: |
...
I could be wrong, usually am actually |
Well, you're not wrong this time!
Of course I completely agree - I didn't mean that it should be done on any circumstances, I just said it's *technically possible*. You stated "no can do" and that's innacurate (you know how ppl take us literally here), so I simply noted - don't say "impossible", just say "bad design".
Geez, this is like some bad teenagre story, he-said and then like I-said, and then...! HE, like, said!  |
|
| Back to top |
|
|
| cactus |
| Posted: Thu Sep 29, 2005 7:01 am Post subject: |
|
|
Apprentice
Joined: 12 Oct 2002
Posts: 30
|
| Quote: |
Stand by is an a way to implement failover, thus you are implementing failover; if you wish you can call it "backup". Regardless of what it's called, this is a highly improper use of MQ, expect many other problems in the years to come, some immediate, others more subtle...
I wonder, what is the buisness need that drove you to such a configuration? |
Right now I can not change the config(Budget doesn't permit..etc) .Down the line we are moving into hardware cluster.
I understand its bad config but I'm not getting why the GATEWAYtoCLONE chl to A(Prod) goes to retry.
_________________
thankx
cactus |
|
| Back to top |
|
|
| wschutz |
| Posted: Thu Sep 29, 2005 7:42 am Post subject: |
|
|
Jedi Knight
Joined: 02 Jun 2005
Posts: 3316
Location: IBM (retired)
|
Interesting problem... tcp/ip rc=32 is "Broken pipe. " Perhaps you've said, but what platforms are involved?
_________________
-wayne |
|
| Back to top |
|
|
|
|
