DataStage Release Level et al
DataStage Release Level et al
This is a topic for an orphaned message.
Fair comment Rick.
We are currently using 3.5r2. I would very much like to go to 4.0 as there is new functionality which would make some tasks I have to do (or mostly have already done) easier (multiple rows from a lookup pops to mind).
We currently have very tight deadlines and budget. We do not currently have a separate environment which would allow us to install and verify that our system still works fine with 4.0 (although I dont anticipate any problems). It will take time and effort to properly manage the installation and verification of 4.0. As DataStage is part of the behind the scenes infrastructure, the users would see no benefit from the upgrade. They have more important tasks for us to spend our time on.
The old axiom of "If it aint broke, dont fix it" also pops to mind. Not withstanding the problem described in my previous message, we have found DataStage to be very reliable. Regarding that problem, I have not yet seen any indication that the problem is properly understood much less actually fixed in any future release. The reply from Muthusamy is interesting (but I note still does not fix the problem) and I too would like to hear more about it. If you could identify a TAR or patch number, that would be useful.
Brad mentioned cleaning up the job. We have not even had to go that far. Just stopping, resetting and rerunning the job is enough to clear the problem (but of course all our jobs run unattended over night). We have certainly not had to reboot the server to clear this problem.
Brad also mentioned keeping &PH& clear to help with performance. I support that comment 100%. I have seen jobs take 20 minutes from when the last stage finished until the job as a whole was finished. We now clear &PH& every night just before the overnight runs start, and there are still 100+ files there the next morning. It would seem to me (correct me if I am wrong, you guys/gals from Informix) that at the end of a job, the controlling process for the job scans through &PH& looking at every file to see if there are any from phantoms that it started. Actually, I think it just looks at the job number in the file as I have seen jobs pick these work files from previous failed runs of a job. The problem in Windows NT is that if you try to open a locked file, it waits around for a while to see if the lock will be released before it tells you that you cant have it. As such, the time to close down a job is also related to how many jobs are currently running and how many active stages are in those jobs.
This also relates back to the hung jobs I talked about in my previous post. These jobs leave zero length work files lying around which UniVerse seems reluctant to release the locks on (even after the job is dead). These locked files also refuse to be deleted.
I hope you all dont mind me responding to all todays messages in the one message. Id better get back to work now.
Cheers
David
-----Original Message-----
From: Rick Schirm [SMTP:rick@schirms.net]
Sent: Friday, 28 July 2000 1:30
To: informix-datastage@oliver.com
Subject: DataStage Release Level
I have noticed in a number of messages that several people are still back on
the release 3.1 and such and some are just going to upgrade to 3.6. Was just
curious on why the upgrade path is not keeping up to the latest release of
4.0.
I have worked with every release of the product since its birth in
pre-release days in 1996. And with every release out there is significant
improvements over the previous. And these improvements out weigh any bugs
that are coming with the product. which is getting fewer.
Not trying to be hostile at all just trying to gather information as to why
people are not up on the latest release available to them.
Thanks for your input in advance.
*************************************************************
Rick Schirm
Application Performart Ltd.
Vice President of Engineering
mailto:rick.schirm@performart.com
web: www.performart.com
phone: 817.291.6177
*************************************************************
*************************************************************************
This e-mail and any files transmitted with it may be confidential and are intended solely for the use of the individual or entity to whom they are addressed. If you have received this e-mail in
error, please notify the sender by return e-mail, and delete this e-mail from your in-box. Do not copy it to anybody else
*************************************************************************
We are currently using 3.5r2. I would very much like to go to 4.0 as there is new functionality which would make some tasks I have to do (or mostly have already done) easier (multiple rows from a lookup pops to mind).
We currently have very tight deadlines and budget. We do not currently have a separate environment which would allow us to install and verify that our system still works fine with 4.0 (although I dont anticipate any problems). It will take time and effort to properly manage the installation and verification of 4.0. As DataStage is part of the behind the scenes infrastructure, the users would see no benefit from the upgrade. They have more important tasks for us to spend our time on.
The old axiom of "If it aint broke, dont fix it" also pops to mind. Not withstanding the problem described in my previous message, we have found DataStage to be very reliable. Regarding that problem, I have not yet seen any indication that the problem is properly understood much less actually fixed in any future release. The reply from Muthusamy is interesting (but I note still does not fix the problem) and I too would like to hear more about it. If you could identify a TAR or patch number, that would be useful.
Brad mentioned cleaning up the job. We have not even had to go that far. Just stopping, resetting and rerunning the job is enough to clear the problem (but of course all our jobs run unattended over night). We have certainly not had to reboot the server to clear this problem.
Brad also mentioned keeping &PH& clear to help with performance. I support that comment 100%. I have seen jobs take 20 minutes from when the last stage finished until the job as a whole was finished. We now clear &PH& every night just before the overnight runs start, and there are still 100+ files there the next morning. It would seem to me (correct me if I am wrong, you guys/gals from Informix) that at the end of a job, the controlling process for the job scans through &PH& looking at every file to see if there are any from phantoms that it started. Actually, I think it just looks at the job number in the file as I have seen jobs pick these work files from previous failed runs of a job. The problem in Windows NT is that if you try to open a locked file, it waits around for a while to see if the lock will be released before it tells you that you cant have it. As such, the time to close down a job is also related to how many jobs are currently running and how many active stages are in those jobs.
This also relates back to the hung jobs I talked about in my previous post. These jobs leave zero length work files lying around which UniVerse seems reluctant to release the locks on (even after the job is dead). These locked files also refuse to be deleted.
I hope you all dont mind me responding to all todays messages in the one message. Id better get back to work now.
Cheers
David
-----Original Message-----
From: Rick Schirm [SMTP:rick@schirms.net]
Sent: Friday, 28 July 2000 1:30
To: informix-datastage@oliver.com
Subject: DataStage Release Level
I have noticed in a number of messages that several people are still back on
the release 3.1 and such and some are just going to upgrade to 3.6. Was just
curious on why the upgrade path is not keeping up to the latest release of
4.0.
I have worked with every release of the product since its birth in
pre-release days in 1996. And with every release out there is significant
improvements over the previous. And these improvements out weigh any bugs
that are coming with the product. which is getting fewer.
Not trying to be hostile at all just trying to gather information as to why
people are not up on the latest release available to them.
Thanks for your input in advance.
*************************************************************
Rick Schirm
Application Performart Ltd.
Vice President of Engineering
mailto:rick.schirm@performart.com
web: www.performart.com
phone: 817.291.6177
*************************************************************
*************************************************************************
This e-mail and any files transmitted with it may be confidential and are intended solely for the use of the individual or entity to whom they are addressed. If you have received this e-mail in
error, please notify the sender by return e-mail, and delete this e-mail from your in-box. Do not copy it to anybody else
*************************************************************************