AIX SRC Progress OpenEdge 10 mkssys Setup.

drobilla

drobilla

New Member
Hello everyone,

We've juste migrated from Progress 9.1D09 on AIX 5.2 to Progress OpenEdge Release 10.1C01 on AIX 5.3.

Everything looks good on the database side. But I'm having a problem to automatically start the OpenEdge daemons when AIX starts up. The old setup was starting everything via a shell script (installed by another admin who no longer works here of course :) I'd like to control the OpenEdge daemons via AIX's SRC commands (i.e. lssrc, startsrc, stopsrc and mkssys).

I can start the daemons with the following command:

su - progress
/home/progress/dlc10/bin/_mprosrv $DBDIR -S 2095 -H $HOSTNAME -N TCP -L 20000 -n 10

But I can't seem to get the appropriate mkssys command to install OpenEdge in AIX's SRC commands. Can anyone help me with this?

Many thanks!

David
 
So far I haven't tried anything because the new machine is undergoing some tests by the users to see if we can go live with it.

Next week I'll have the system available for this. Until then, it is assumed that we must manually start the OpenEdge services.

But another systems admin told me he did try it with AIX 5.2 and Progress 9. He said suggested not to use AIX's SRC system for progress as it caused him some problems when the database wouldn't start, that would hang the machine's boot process. Any heard of this?

Thanks,

David
 
drobilla said:
So far I haven't tried anything because the new machine is undergoing some tests by the users to see if we can go live with it.

Next week I'll have the system available for this. Until then, it is assumed that we must manually start the OpenEdge services.

But another systems admin told me he did try it with AIX 5.2 and Progress 9. He said suggested not to use AIX's SRC system for progress as it caused him some problems when the database wouldn't start, that would hang the machine's boot process. Any heard of this?


David
Click to expand...

I wouldn't recommend using AIX src for Progress under any circumstances. AIX 5.3 has cavets that are unique to that version of UNIX (shared memory limits in particular)

Most importantly, if the system is rebooted without warning, or if the system encounters when the system is brought down.

If you attempt to start a database that has been corrupted due to a hardware failure as part of the boot, that could further damage the database or prevent adequate evaluation of the level or corruption.

---

UNIX systems shouldn't come down much, if at all. The rarity would lead me to believe that startup of any Progress process should occur outside the realm of the boot process.
 
> If you attempt to start a database that has been corrupted due to a hardware
> failure as part of the boot, that could further damage the database or prevent
> adequate evaluation of the level or corruption.

Well, I'm used to Oracle databases which are highly corruption resilient. In 10 years I've lived through too much machine crashes but Oracle has always come back ok. I'm not that familiar with OpenEdge/Progress, so I'll take your word for it :)

> UNIX systems shouldn't come down much, if at all. The rarity would lead me to
> believe that startup of any Progress process should occur outside the realm of
> the boot process.

I totally agree that UNIX machines don't go down very often. You still have to reboot them a couple of times a year to keep up with Security patches and bug fixes.

However I don't agree that something should be started manually. To me that's one more thing you can forget. We you run serveral hundred machines, you don't want to have things to remember like this. That's why everything else I run starts automagically when the machines go up.

Anyway, it seems OpenEdge is a different animal and since I only have two machines running it, I guess I'll just keep on starting the database manually :S

Thanks for your imput.

Cheers,

David
 
You could do that or start them as a CRON job. Evaluate if they are running and attempt to start them if they aren't.
 
OpenEdge is at least as uncorruptible as Oracle.

Personally I don't know anything about AIX "src"commands but you can easily start a db through the usual UNIX /etc/rc.d startup process on any UNIX and many people, especially those with many machines and many databases, do so.

Of course any automated process like this needs to handle exceptions properly and let you know if something is wrong -- 99.44% of the time a crashed db is just going to go through crash recovery and restart without any issues. Especially if the db crash was a result of a server crash. The exceptions are going to be things like a filesystem holding part of the db has disappeared etc -- those obviously need manual intervention.

In my experience people who damage their db with automated startups have done so by doing things that over-ride the safe-guards that Progress has built into the product. Not because of any inherent flaw in Progress.
 
I suppose I may be a little paranoid, but upon a system startup I would rather script the startup of all systems using ssh remote commands from a single system manually.

I would rather take 10 minutes out of my day to start a process and view it run after a system outage the auto-start my database; a risk of .56% is still a risk:

- If I had a crash, my butt would be at the office no matter how many safeguards I think I may have.

- If there is a reboot planned, I am making the assumption kernel changes/updates/patches were installed; my system admin should have the ability to take corrective action, were there a problem on reboot with any of the patches.

But if the environment is unstable, I guess I would argue the case that automatic startup is worth the risk.
 
Different strokes for different folks.

BTW -- I made up the 99.44% out of thin air. The point is that there is rarely any problem at all. And when there are problems they are not of the "corrupt the database just by starting it" variety. You have to go out of your way and do something ill-advised to corrupt a Progress db. Things like using -F in an automated script...
 
> OpenEdge is at least as uncorruptible as Oracle.

Hey, no offense intended! :) Like I said, I'm new to OpenEdge. While I have many years of Oracle experience, it doesn't mean I think it's better/worse/equal to OpenEdge. To me it's just nice to know another way of doing essentially the same thing. New client, new software, new ideas. Great!

> Personally I don't know anything about AIX "src"commands but you can easily
> start a db through the usual UNIX /etc/rc.d startup process on any UNIX and
> many people, especially those with many machines and many databases, do so.

Of course. But I tend to integrate as much as possible de software with the particular OS that's running on. So I've written differents scripts for RedHat, Ubuntu & FreeBSD because these OS require rc-style scripts. I've done Service Management Facility (SMF) stuff for Solaris and now I'm trying to do System Resource Controller (SRC) stuff for AIX. That's all.

> Different strokes for different folks.

Damn right!

Cheers guys and thanks for your imput,

David
 
You must log in or register to reply here.
Back
Top