My client has a computing environment that includes six F630's, an F540, and an F330, all doing CIFS, NFS, and quotas. In January, we decided to get them all running the same OS revision, and based on the bug list we chose 5.2.1 as a target.
So we upgraded one of the F630's to 5.2.1 in mid-January to see how it'd do. It did great, so Friday before last we upgraded the remainder of the filers from various 5.1.x flavors to 5.2.1.
At midnight on Saturday night/Sunday morning, all but one of the filers panicked. The F630 that had been up for a month on 5.2.1 joined the group in panicking, but an identical F630, just upgraded the previous day with the rest of the group, did not.
The next day, NAC technical support told us, even before they saw the dump, that this was bug 12363 ("Terminated DC connection causes crash in stub code") which is new to 5.2.1 and occurs when a CIFS-enabled filer loses contact with its domain controller. This bug is fixed in 5.2.1P1. We were skeptical of this diagnosis at first, since we hadn't had any known network events or DC events at that time, but conveniently a filer panicked again on Monday when we tried to delete a share. That data point plus NAC's report on Sunday's dumps, which confirmed their diagnosis of bug 12363, convinced us to begin a careful yet urgent move to 5.2.1P1.
We installed 5.2.1P1 on one of the F630's that had panicked. No problems for the rest of the week. So Sunday we installed 5.2.1P1 on the rest of them.
Two filers (one F630 and the F330) panicked Monday after we did a "quota on" on them. The panics occurred on the same failed assertion: "../common/wafl/quota.c:926: Assertion failure." (Because the "quota on" is part of a script that various folks run frequently, it took two panics before we drew a straight line between the "quota on" and the panic.)
The panics recurred on attempt to reboot. We had to boot off floppy, leaving quotas off.
So now we are running without quotas on the filers that panicked on 5.2.1P1. And we can't do a "quota on" on any of the others; I personally am even afraid of "quota resize". And we are faced with the difficult decision between regressing to 5.1.x (which wasn't a rose garden either) or waiting for a new 5.2.1Px. Either way, we've submitted the most recent dumps to NAC tech support, and we're waiting on a response.
If you are thinking of moving to 5.2.1, I encourage you to think carefully.
Brian
My client has a computing environment that includes six F630's, an F540, and an F330, all doing CIFS, NFS, and quotas. In January, we decided to get them all running the same OS revision, and based on the bug list we chose 5.2.1 as a target.
If you are thinking of moving to 5.2.1, I encourage you to think carefully.
I will second that. We upgraded our 630 to 5.2.1 on Friday, and it hung this morning. Color me not thrilled. That said, I wouldn't expect Netapp to have magically broken the "sawtooth bug curve" somehow, so I guess we should have simply waited for 5.2.1something.
Paul
If you are thinking of moving to 5.2.1, I encourage you to think carefully.
Our encounter (albiet brief) with 5.2.1 was pretty disasterous too. I'm camped at 5.1.2P2 and have no reason to move. Except for the occasional 5.2.1 floppy boot for the fiber channel diagnostics. ;-)
One day, i'd really like to see NetApp move to a user patchable OS. Anyone else?
Graham
I'm very grateful to read this. After about six OS upgrades ( mostly debug releases), I stuck at 5.1.2 and have refused to move from that, despite Netapp's assurances that the next OS was more stable or whatever. Over a five month period, I had the joy of changing OS on my F520 no less than six times. Compare that with the Auspex I had for three years prior to that. On the Auspex, I had to upgrade the OS once ( and it didn't involve a reboot). In my painful experience, NetApps should have better Quality Checking and Testing before they churn out a release. Anyone have any idea how many NetApps OS's there are in use ?
Raymond
-----Original Message----- From: owner-toasters@mathworks.com [mailto:owner-toasters@mathworks.com]On Behalf Of Graham Knight Sent: Tuesday, March 02, 1999 10:39 PM To: Brian Rice Cc: toasters@mathworks.com Subject: Re: 5.2.1 blues
If you are thinking of moving to 5.2.1, I encourage you to think
carefully.
Our encounter (albiet brief) with 5.2.1 was pretty disasterous too. I'm camped at 5.1.2P2 and have no reason to move. Except for the occasional 5.2.1 floppy boot for the fiber channel diagnostics. ;-)
One day, i'd really like to see NetApp move to a user patchable OS. Anyone else?
Graham