System Hangs

[Keith Parris]

Availability Manager (or DECamds) is a good way to see what's going on within a system during a "hang". This comes free with a VMS license on Alpha or VAX and there's even a kit on the OS media.

The Data Analyzer for AM can run on either a PC or a VMS box; The earlier DECamds software runs only on a VMS box. There's a Data Provider (same for either AM or DECamds) that is installed on the box to be monitored (it shows up as RMDRIVER).

Since queries come in over the LAN at IPL 8, unless the box is locked up tight at hardware interrupt level or something, you can see what's going on even if you can't log in interactively. See http://h71000.www7.hp.com/openvms/products/availman/index.html

[David B Sneddon]

Mike,

How intermittent are the hangs?
Is it possible something is using up all your
pagefile space?
As Keith suggests, DECamds can be very useful
for these types of situations.

Regards
Dave

[Wim Van den Wyngaert]

I am always amazed of how easy everyone is doing "upgrades" without re-qualification of the applications. I freeze my systems the moment the qualification is done.
And no one-time, uninvestigated system hang will motivate an upgrade.

Wim

[Volker Halle]

Mike,

using DECamds/Availability Manager also allows you to force a crash from remote (without the need to change motherboard jumpers), if the system hangs (assuming it does not hang at or above IPL 8).

You just need to configure the correct security in AMDS$DRIVER_ACCESS.DAT to allow WRITE access from your AMDS Data Collector node.

If you keep an AMDS Collector node running all the time (until the next system hang), you can also look at the AMDS Event messages (AMDS$LOG:AMDS$*.LOG) prior to the hang. If the system hangs, try to look for problems using AMDS, then force a crash (to capture the dump and get the system back up) and then you can take your time to look at the problem in the dump. This will also cause error log entries to be written to ERRLOG.SYS.

Volker.

[Mohamed K Ahmed]

In response to Wim's reply,
Most applications running on old versions of OpenVMS would have been upgraded to accomodate the new versions. If not, then you have a problem with the applications and should seek advice from the vendor.

In this situation, he has version 6.2, which is very old in contrast to the new versions, so an upgrade would be the most appropriate thing to do (of course after checking the applications).

Mohamed

[Michael LaRoche]

Upgrading th OS was of first interest to me but the decision makers are saying lets wait until we get the new hardware and applications then all we have to do is move the database files and update them on the new system.

I also wanted to put all the patches in to make it up to date but again and according to the powers that be it would take over a year and a half to put them in piecemeal, which is the desired way here nothing all at once just a couple at a time.

[Wim Van den Wyngaert]

Mohamed,

The real price of testing applications is underestimated. To test a big application properly weeks if not months are needed. And a new machine because otherwise you don't have the 2 versions available for development.

And if you don't take this time, you are taking risks. And that can cost even more (things not working after the upgrade).

6.2 may be old but we recently had machines running it being up for years.

Wim (not upgrading anything)

[Antoniov.]

Hi Mike,
I'm using alphaserver 400 4/233 with VMS V6.2 and with V7.3-2.
Lat year, one machine hanged intermitently or could not boot. After some weeks crashed. I changed power supply and now it works fine!
Last week, another machine broken power supply!
May be you have similar trouble.

Antonio Vigliotti

[Michael LaRoche]

Will need to look at it when it happens again. So far no users have been affected, happens at night or weekends when no one is on, and some times when either backups are running or just the machine is up and running with no other interaction.