#14 mbsync hangs with "unhandled SSL error 6"

[Tomas Tomecek]

isync started to hang on a pretty big folder:

$ mbsync <account>

Selecting master folder/folder...
Selecting slave folder/folder...
Loading master...
Loading slave...
slave: 9834 messages, 0 recent
Warning: lost track of 80283 pulled message(s)
master: 90117 messages, 0 recent
Synchronizing...
M: +0/0 *0/0 #0/0  S: +16770/80283 *0/0 #0/0
Socket error: secure read from <mail_server>: unhandled SSL error 6

When I rerun the command, error message tends to change:

Socket error from <mail_server>: Connection reset by peer

Socket error from <mail_server>: Broken pipe

Since this is our company mailserver, I don't think that problem is on the server side.

I've tried to use latest version from git master and the result was the same.

I can supply whatever traces/logs are needed.

[Jeff Peeler]

I'm getting this error as well using version 1.1.1 on Fedora. Is there anything that can be done to further this along?

[Arne]

I am getting this too with the latest version from git. This started happening after I upgraded today. "mbsync -a -l -V -D": http://pastebin.com/nSh18nQe

New messages do not show up in my inbox. My inbox is not very big, it has around 250 messages.

[Oswald Buddenhagen]

the OP's output indicates that the connection "simply" breaks down. this happens, especially with company mail servers ... because they tend to use m$ exchange, and that one sucks as an imap server.

arne's log indicates that connection shutdown isn't handled properly, but that's about it what i can tell. it would be interesting to know how reliably the problem happens always at the same place.

for debugging, an strace log would be probably somewhat helpful, but i don't know. with some quick googling i couldn't even find out what that "error 6" is supposed to be.

[Tomas Tomecek]

Thanks Oswald,

I've opened an issue with our IT dept.

BTW, we use zimbra.

[Arne]

This behaviour is consistent after updating. I have two computers side by side on the same network connection. One works every time, the other does not work at all. The config is identical, i keep my dotfiles in a git repo.

The one that works: (Built from commit: f377e7b)
$ mbsync --version
isync 1.2.0

The one that does not work: (Built from commit: 74c78c7)
$ mbsync --version
isync 1.2.0

I might try to bisect the issue later today or tomorrow.

[Arne]

I did a git bisect this morning, it seems the problem that I am having was introduced in commit 6c959c3ee4bba035e6ad0e8e71cb7526f503e43b. I hope this helps in tracking down the problem.

[Oswald Buddenhagen]

well, you're the second one to bisect it to that commit, but that makes no sense - this commit is from dec 13 (and entered master just a few days ago), while the original report is two months older than the commit.

so i'm assuming that the underlying problem is unrelated to the commmit, only that your server/network/whatever in conjunction with that commit triggers it.

[ScottM]

For what it is worth, I get this same error 100% of the time with mbsync/isync 1.2.0 on OS X with Fastmail as the server. I doesn't hang, per se, it merely reports the same SSL error on terminating the connection:

6 LOGOUT
* BYE LOGOUT received
6 OK Completed
Socket error: secure read from mail.messagingengine.com (66.111.4.51:993): unhandled SSL error 6
Segmentation fault: 11

If it matters, mbsync is using:

/usr/local/BerkeleyDB.6.1/lib/libdb-6.1.dylib (compatibility version 0.0.0, current version 0.0.0)
/usr/lib/libssl.0.9.8.dylib (compatibility version 0.9.8, current version 0.9.8)
/usr/lib/libcrypto.0.9.8.dylib (compatibility version 0.9.8, current version 0.9.8)
/usr/lib/libSystem.B.dylib (compatibility version 1.0.0, current version 1197.1.1)
/usr/lib/libsasl2.2.dylib (compatibility version 3.0.0, current version 3.15.0)
/usr/lib/libz.1.dylib (compatibility version 1.0.0, current version 1.2.5)

[Oswald Buddenhagen]

can you make a backtrace or valgrind trace of that segfault?