Executive summary for those who don’t want to read anything beyond the bare details: On Saturday, August 23rd, both the MCS Mail & Calendar (aka Zimbra) and the MCS Production network will be undergoing maintenance and will be unavailable from 6AM until 6PM. Also, on Sunday, August 24th, Zimbra will be offline from 8:30 to 10:30 AM.For details, read on. Read the rest of this entry »
Hey, folks. Hope everyone’s having a great summer. I was going to put this note into the next Dispatch from the Geeks, but I’ve had some requests on the matter recently, so I figure I’d send along this note right now. Today’s lesson comes in the form of a parable of the Happy Little Bug.There once was a happy little bug. His name was Desmond. Desmond, the happy little bug.Desmond sent large print jobs to the printer and then never went to check on them. Then the printer ran out of paper. Then all the other bugs who went to use the printer had to load it with paper but then had to wait a long time for their jobs to come out because of Desmond.So they squashed him with a rock.The end.The moral of the story: if you’re sending a big job to the printer, make sure you check on it to make sure it came out okay. If you’re using a Mac or Windows machine, you can send the job with a password so it won’t print until you get there. See http://mcs.anl.gov/faq/Printers for the latest drivers with this capability.Thanks!
If you use iCal and sync your calendar with Zimbra, or would like to start, I’ve posted detailed instructions on how you can do this with CalDAV here:http://wiki.mcs.anl.gov/IT/index.php/Mail#Zimbra_and_iCalI’ve seen enough failures and sync issues with the Zimbra iSync connector that I’ve lost faith in it for Calendar syncing. It’s okay for Contacts syncing between Zimbra and Address Book (and is the only way to accomplish that right now), but I’ve had enough users complain about appointments not making it up to the server or not make it from the server to their own calendar that I just cannot in good faith recommend using it anymore.But this new method requires running OS X Leopard (10.5). Tiger users will have to continue to use the connector.Now, that being said, there’s a small change in how you interact with your calendars, especially iPhone users (though iPhone 2.0 should pretty much solve that). However, the difference in reliability, in my opinion, more than outweighs the burden of slightly changing your operating mode. This is, by far, the most reliable way to make sure the server is up to date short of just switching to the web interface exclusively. The details on the changes are on the wiki page above, but the short story is that every calendar you have on your Zimbra account, including shared ones, will be synced to iCal, and by default all alarms on all calendars are turned on, though you can turn them off individually or all at once.If you have others who manage your calendars or regularly rely on the server view being up to date, you should strongly consider switching to this mode of operating. No more silent failures, and instant updating of your appointments to the server when you enter them on your local calendar. Plus, I’m told it will also help prevent hair loss and random cranial explosions.If anyone wants any help setting this up, I or someone else in the group should be able to assist, though it’s not that difficult. Also, if you have suggestions on improving the documentation, please pass them along.Thanks!
The dry details: Zimbra upgrade this weekend is complete. It seems to be faster and the issues we’d seen with dropped IMAP connections and blank e-mail copies seems to be solved, let us know if this is otherwise. There are new features in the new server, which you can explore at your leisure, and iCal sync instructions have been updated.The pointless exposition, in which I show my age something fierce:Come and listen to a story ’bout thing called mail,it went through a big change, but still kinda failed.Until finally, the decision got made:It was once again time to push through a ‘grade.Upgrade, that is.Five-oh-six.And faster disk.Of course, the next thing you know, it didn’t go as planned:took a couple hours more, but things were well in hand,And ’round about 3, things were lookin’ pretty fine10 minutes more and we were finally online.Back on-line!New features!Bug fixes!Yes, folks, the upgrades this past weekend appear, thus far, to be a success. There were some big issues we were hoping to see resolved, and it feels like they have been. Here’s what we’d noticed over the past few weeks:
- Dropped IMAP connections: Users were experiencing drops of their IMAP connections with a certain regularity. Especially overnight. We believe this issue has been fixed based off of preliminary reports. I should note, though, that everyone’s IMAP connection will drop one night per week when a backup is run on their account. When your account is backed up, all connections to it are killed and the account is set to maintenance mode for the duration of the backup. The accounts are divided into seven groups, and one group is backed up in full each night. (Redo logs cover any changes between backups, so we can restore with a much finer granularity than just one backup per week).
- Slow response: Web and mail clients would see laggy response logging in. This should also be addressed, though I’ll note that Safari 3 and Zimbra 5 interoperability is considered ‘beta’ by Zimbra at this stage. They are still working on several known issues.
- Blank e-mail copies: Some users (myself included) were finding large messages with attachments would result in blank copies appearing when those messages were copied via a server-side filter. This bug should be no more.
- Redlivery of already-delivered mail: Likewise, if a user was forwarding their mail, and the message was large, the user would see repeated attempts to deliver the mail as long as the forward rule kept failing on large messages. This appears to be fixed.
On top of that, we’ve got us some new features, specifically:
- Scheduled Out of Office Autoreplies: When you set your Out of Office message, you’ll note that you now have the option to specify a start and stop date for the messages.
- Flexible alarm schedules: Zimbra is now supporting flexible alarm schedules on a per-appointment basis via the web client. (Why it took them this long, I’ll never know.)
Also, for you iCal users out there running Leopard, you can find instructions on how to sync your calendars without using the iCal sync connector at http://www.mcs.anl.gov/faq/Mail#Zimbra_and_iCal. It’s been reliable and solid for me, though it’s not as cut and dry to set up. Also, it’s a little inconvenient for iPhone users who regularly add appointments via their phone, though I’m hoping that clears up once ActiveSync lands in July.Last tidbit, the Web version of this blog is now iPhone friendly.That’s about it for the post-mortem on the upgrade. And thus…Now it’s time to say ‘bye in this Dispatch from the geeks,I’m sure there’ll be another one within a couple weeks.You’re encouraged to take a peek at our friendly, helpful logat MCS dot ANL dot gov slash systemsblogWeb 2.0!Subscribe to RSS!Sit & read a spell!Y’all come back now, y’hear?And yes, I just made a valid mcs.anl.gov/systemsblog redirector so it would fit with the meter of the lyrics. Sue me.
Took a little longer, but we’re back. We’re still monitoring things, but everything looks normal. I’m going to run some tests on some known bugs we wanted to fix, and I’ll post an update later with some of the new features you can now use.Thanks for your patience, all.
Hit a small snag. It shouldn’t be much longer, probably another 60 to 90 minutes. See the CIS outage page for details.
Short status update for those who may not have seen the notices or forgotten, the mail server upgrade is under way. Things should be back by 1:00 PM Central. Stay tuned.
First up, some of you would have noticed an issue on 6/11 getting to Zimbra. The very short version of the story is that in anticipation of our existing SSL certificate expiring, a new one was ordered. Unbeknownst to anyone here, when they issued the new certificate, the existing one was revoked.Anything that uses OCSP to check the validity of certificates immediately noticed this. Firefox 3 has this enabled by default (and, really, that’s a good thing), so those users all felt the pain.We got the new certificate installed, and all is now well, so if you disabled OCSP as a workaround, you should re-enable it.We also had a problem with the certificate for sending mail, but it only seemed to affect Outlook users. We’ve isolated that one and things will be back to normal shortly.Next up, the aborted upgrade this past Saturday. Around 11PM on Friday night, we discovered an issue our planning and testing didn’t cover. At that point we had three options:
- Wing it and go ahead with the upgrade.
- Temporarily push the upgrade back a week or two while we figured it out.
- Do the first part of the upgrade (moving to the latest server version), but not the second part (moving the service to a 64 bit OS with faster disk) until the next major outage window.
We decided to sleep on it. At 8:45 on Saturday morning, we made the decision to abort and wait at least a week, with the understanding that if we could not figure out the second part of the upgrade, we’d still do the first part, since we know that will at least fix some things.The solution to the issue has been found, and we’ll make the upgrade time official once testing is complete. We currently expect the same outage window we had last weekend (9AM – 1PM Saturday, June 14), but an official announcement will come soon.Thanks for your patience.
We’re postponing the upgrade to the Zimbra server. In the run-up to the migration, we discovered an issue that we had not previously identified in the plan. It’s not a big issue, mind you, but rather than improvise something, we’re going to make sure this update goes as smoothly as possible. So, we’re going to take another week to think about it and submit a new upgrade plan to the Change Advisory Board.Mail operation will continue as normal today.Sorry for the inconvenience.
It looks like we’ll be able to go ahead with the server upgrade I sent mail about earlier this week. You may get another notice from CIS on this, but here is the current plan:
- Upgrade to ZCS (Zimbra Collaboration Suite) software v5.0.6 from 5.0.2 (bugs fixed/features improved)
- Upgrade from 32bit OS to 64bit (improved capacity and performance)
- Upgrade to Tier 1 high speed disk (improved capacity and performance)
The maintenance window for this procedure will begin at 9:00AM this Saturday morning (2008.06.07). We anticipate completing the upgrade process by 1:00PM (same day).During this period, the Zimbra server will be unreachable. You will not be able to check your mail, calendar, or anything else that is stored on the server. Incoming mail will be held and delivered when the upgrade is complete; no mail will be lost.Upon completion, we’ll send an “all clear” message with any pertinent details.I will post updates on the process as they’re available at here. Please let us know if you have any questions.Thanks!