Outlook and Cyrus (or any other IMAP server)

We've recently had the "pleasure" of investigating Outlook to make it work better with Cyrus when using IMAP, but as it's most about Outlook, this might help for all you dovecot and other users as well.

• There's no escape from PST hell! Your PST files need to be undamaged and preferably local for Outlook to work good. We haven't investigated any of the non-MS repair tools, but scanpst.exe is a good start.
• The default configuration for Outlook is "not so good" - it will traverse the entire folder structure regularly. That wouldn't be so bad except that it uses a rather short (fixed) timeout for the entire operation, and complains loudly and behaves badly if doesn't complete in time. Under Tools > Send/Receive > Send/Receive Settings > Define Send/Receive Groups -> All Accounts you should deselect anything but the Inbox.
  You'll also need to deselect "Get folder unread count" or it will still traverse your entire folder structure.
• Though the need seems to vary, I'd still recommend running with some hack to wake up Outlook from IDLE in order for the connection not to time out and be disconnected. Dovecot seems to contain such a hack natively. We're on Cyrus and I've whipped out a hack that does the same thing. (Lie to the client saying we have new mail, immediately followed by an expunge and new information where that faked mail has disappeared).

The second bullet is particularly exciting to me - for starters, I've strongly preferred if the timeout was per sub action (actual IMAP command) as that wouldn't break things for those who use a large amount of folders.

I was also rather thrilled to discover that "Get folder unread count" automatically recurses and queries all subfolder of the selected folder, without giving the user any hint about what's happening.

Migrated

After having successfully migrated Himmelmora mo to blogger, I quickly decied to migrate Sky heath as well. As you'll probably have noticed some way if you're reading this.

Large data areas - how much can you malloc

One of the funny (in a tear-your-hair fashion) things with AIX is that it has so many process models, and no matter what they will give you trouble somehow.

One common issue for us is that we hit the default limit of 256 Mbyte data per process. And of course, our applications usually don't signal that they run out of memory, instead we get weird segmentation faults and other freaky stuff.

And of course, these issues do not surface often enough for this to be the first thing we think about. But everytime it does it makes me want to set MAXDATA for everything (which I of course won't do).

Modernized

Thanks to Google reader, I finally got around to fix an atom feed (a bit slow given that RFC 4287 was published in December).

Anyway, it seemed a bit weird to me that no one had fixed an atom feed for my system (COREblog 1.x) before. And then I realized how outdated my Zope skills (and software) were - I guess I really should learn ZPT some day, but it works well enough for now.

I guess I'm a junkie

Congratulations the world. ISO 26300 is now available.

Unfortunately, we're still using some software not up to standard, but we're working on it.

What makes software great

I intended to whine about OS X and bundles, but as one of our mail servers dumped earlier today, I'll write about that instead.

One of the softwares we put a great deal in is IBM SAN FS. As is it a proprietary, closed product, I don't have a clue as to the code quality, or the development model. Or lots of other things that influence the quality of the overall product.

What I do know is that the support team is truly excellent, which compensates for the closedness of the product, something very rare in my book.

We've had (and continue to have) some issues, but no matter what cluster file system we'd chosen, I'm rather convinced we'd be a sort of test lab, and while I'd personally prefer at least being able to find and fix errors ourselves, paying someone who knows the code base to fix errors is a much smarter thing to do when trying ourselves (and this is the same for open and closed products of this magnitude).

TSM

I really do like TSM as a backup system. But the code really seems to be something to scare little kids with. I remember the first time I tried running it on a freshly installed Debian sarge. I got it working eventually, but it took rather long.

Of course, TSM is the only PPC-program I've used so far that just doesn't work on an Mac mini (intel):


pontus:~ pontus$ sudo dsmc incr
IBM Tivoli Storage Manager
Command Line Backup/Archive Client Interface
Client Version 5, Release 3, Level 3.0
Client date/time: 2006-05-02 14:10:13
(c) Copyright by IBM Corporation and other(s) 1990, 2006. All Rights Reserved.

Node Name: PONTUSMINI
Session established with server TSM2: AIX-RS/6000
Server Version 5, Release 3, Level 2.2
Server date/time: 2006-05-02 14:10:11 Last access: 2006-05-02 14:10:08

Incremental backup of volume '/'
dsmc(6146,0x82015800) malloc: *** vm_allocate(size=2869547008) failed (error code=3)
dsmc(6146,0x82015800) malloc: *** error: can't allocate region
dsmc(6146,0x82015800) malloc: *** set a breakpoint in szone_error to debug
dsmc(6146,0x82015800) malloc: *** vm_allocate(size=2869547008) failed (error code=3)
dsmc(6146,0x82015800) malloc: *** error: can't allocate region
dsmc(6146,0x82015800) malloc: *** set a breakpoint in szone_error to debug
dsmc(6146,0x82015800) malloc: *** vm_allocate(size=2869547008) failed (error code=3)
dsmc(6146,0x82015800) malloc: *** error: can't allocate region
dsmc(6146,0x82015800) malloc: *** set a breakpoint in szone_error to debug
dsmc(6146,0x82015800) malloc: *** vm_allocate(size=2869547008) failed (error code=3)
dsmc(6146,0x82015800) malloc: *** error: can't allocate region
dsmc(6146,0x82015800) malloc: *** set a breakpoint in szone_error to debug
dsmc(6146,0x82015800) malloc: *** vm_allocate(size=2869547008) failed (error code=3)
dsmc(6146,0x82015800) malloc: *** error: can't allocate region
dsmc(6146,0x82015800) malloc: *** set a breakpoint in szone_error to debug
dsmc(6146,0x82015800) malloc: *** vm_allocate(size=2869547008) failed (error code=3)
dsmc(6146,0x82015800) malloc: *** error: can't allocate region
dsmc(6146,0x82015800) malloc: *** set a breakpoint in szone_error to debug
dsmc(6146,0x82015800) malloc: *** vm_allocate(size=2869547008) failed (error code=3)
dsmc(6146,0x82015800) malloc: *** error: can't allocate region
dsmc(6146,0x82015800) malloc: *** set a breakpoint in szone_error to debug
dsmc(6146,0x82015800) malloc: *** vm_allocate(size=2869547008) failed (error code=3)
dsmc(6146,0x82015800) malloc: *** error: can't allocate region
dsmc(6146,0x82015800) malloc: *** set a breakpoint in szone_error to debug
Abort trap
pontus:~ pontus$