spond's blog

Do you ever have one of those weeks... followed by another and another? My main NetWare server in my lab has been suffering hardware problems recently - the motherboard decided to die (this I discovered after testing a different processor and power supply). Luckily my brother-in-law had an old PC with a similar motherboard that he wasn't using, so I used that, which also gave me a nice processor upgrade [not that NetWare needs much in the way of processor, but it helps the startup to run quicker :-) ]. This process had taken me about a week all told, and I was getting nervous, because this is also my mail server, and I was starting to lose mail.
So when it all powered up, I was happy, mail from the last 4 days started to trickle in and I thought I'd found a nice low-cost way to prolong the life of this ancient (more than 6 years old, like I said, ancient) machine. Little did I know that there was something else lying in wait - a couple of days ago, I rebooted the server after making some changes and... it wouldn't boot up (same symptoms as before, power to the motherboard and disks, but no video and no "beeps"). Draining the CMOS allowed it to partially boot - once, then not again. Tried 3 different power supplies, nothing helped (my theory at this point is that my original power supply may be bad, and it fried the motherboard - well both of them actually). So I was feeling a little despondent as I'd wasted a lot of time on this. I decided to bite the bullet, and buy a new server, 5 minutes online with Dell and a server was ordered, but of course there's a 2 week waiting time, what to do meanwhile...

One of my friends, Andy, reminded me that VMWARE has the facility to use a physical disk... So, taking the disk, plugging it into my main Linux box (SLES9 at the moment), and setting up a new virtual machine, I was all set I thought, NetWare started to boot... and hung at the "Pres Esc to cancel boot" message, but there was disk activity. I left it for one hour, two, and found the box was running at 100% load, nothing would respond. In the end I had to pull the power on the Linux box as it was so unresponsive. I tried this a few times, with different startup parameters, but still it was unusable. I felt soooooo close, but what was causing this? Searching through the disk parameters, I had a Eureka moment - but this time it wasn't in the bath. Change the disk type to be "persistent", so that VMWARE wasn't trying to make copies of everything that was happening. Voila! System boots, it's working just fine now as a virtual server, and when the new box arrives, I'll be doing an across-the-wire migration.

I LOVE virtualisation...

Oh, an update, based on some work I did today - it looks like the problem that caused the slowness was me making the disk the IDE slave:  changed it to master and that gave me back the speed...

ZENworks Patch Management (ZPM) 6.3 is nearly here, and I've been testing it. Looks very good, with most of the things I was hoping for. One thing intrigues me - do you have ZPM, and if so, do you use it for NetWare (if so, why?).

An update - still testing ZPM 6.3, anyone who's looked at Patchlink's site will see that there have been some teething troubles with this release - Novell wants to make sure that the version we release has these bugs fixed. So I've reinstalled my server with 6.2.2.181 from ZEN7SP1 and I'll be testing the upgrade to a new build over the next few days. If it's good, we could be in a position to release it soon (yippee!)

Currently working with Martin Irwin, the CPK king, on an article for Cool Solutions where we will list all known CPKs/SPKs, what they do and where to get them - if you know of any you'd like included, or any info for existing ones, let us know...

I'm full of admiration for people who find time to update their blog every day, perhaps they were inveterate diary writers, and this is now an easier way to do it, with the added ego boost that someone else may read it! And why is it that everyone waits for you to go on vacation before they send their email? I came back from a week in France to 261 emails and 4500 unread forum messages!

Since my last entry, I finished off the update TIDs for ZMD7 and ZDM6.5, and gosh darn it, we've released another patch for 6.5 (in beta) so that needs doing again. Last week I got involved in creating a CPK for one of the hotpatches, and I wondered if it would be a good thing if we created one for all of the hotpatches - we already create one for support packs, but I think that hotpatches deserve it too. I'm going to look at this very soon (I'm away next week with some of the SysOps on their "retreat", but expect this to be a high priority for me when I return) - does anyone think I'm wasting my time?

Spent an interesting (and exhausting) weekend starting the process of pulling together the list of fixes for ZENworks 7 Support Pack 1, so we can publish them. For example TID3484245 is the result of working from about 5am to 3pm both days (I know what you're thinking, why can't he be more efficient? ).

Well the truth is, that although Novell does have a good system for tracking bugs (Bugzilla), not every bug gets a TID written when it needs one. It's very easy when you're dealing with a problem for one customer: you raise a bug for their issue, and when it gets fixed, they're happy, and you move on to the next problem - I've done it myself, though I try not to. What we need to get better at, is deciding whether other customers would benefit from knowing about the problem and the fix. Now that we have a new system for letting customers know about hotpatches, it's more important than ever that we raise TIDs earlier in the process. What's needed at this point is to go back to the engineers who raised the bug reports, and persuade them that writing a TID for an old problem is as important as trying to fix a problem for the customer that's screaming now!

I know that many customers need these lists, to be able to show their boss that the issue they're experiencing will be fixed by the patch: Novell encourages customers to upgrade to Support Packs (after all it's free, isn't it?) but if you've worked in a large corporation, you know that free doesn't mean no cost - there's the testing cycle, the paperwork, the pilot, the rollout, and the assumption from your end users that anything that goes wrong from that point on is because you did the upgrade. You're in trouble if the weather turns bad after your upgrade, because that'll be your fault too!

I'll post updates here as these lists get updated, to give you a feel for how the process is going. I'd be very interested in any comments you may have on the good or bad points regarding this process, or TIDs in general.

July 17 - added a bunch more TIDs today, deleted a couple of duplicates (hey, maybe we fixed the problem twice? ). I'll be adding more tomorrow, then I'll start chasing the recalcitrants...

I'm running some tests now for someone who's having problems with the NAL Explorer failing to run, but he has over 600 app objects associated - what's your record for apps?