Re: Managing IMAP book
On Fri, 11 Feb 2005, Amos wrote: > Aleksandar Milivojevic wrote: > > FC3 and RHEL4 include two IMAP servers, Dovecot and Cyrus. Dovecot > > seems to be the choice for small installations and/or novice admins that > > were using wu-imapd on older distributions. Basically it is drop-in > > replacement for uw-imapd. Cyrus is included for admins that need > > advanced IMAP server. > > Oh, and I forgot that Apple's OS X Server comes with Cyrus for the > IMAP/POP service. > > > Anyhow, I hope that this inclusion will warrant new book. More people > > will probably try out Cyrus now that it comes neatly packaged with > > reasonable default configuration files. Plus, it seems that many people > > do not know that Dovecot is IMAP server ;-) > > If folks are interested in seeing an update, I would recommend sending > such requests to O'Reilly. They certainly won't approve an update if > there's no indication of demand. > > Of course if an update is approved, it would be sad if it came out and > then immediately afterwards 2.3 was released, but then I guess that's > life in the software world. 2.3 should be pretty close to being released. If an update was done now, I think it should be targeted toward 2.3, with the expectation that 2.3 would be out before the book is pressed and the feature set of 2.3 is not going to change drastically. But that's only my opinion. - Bill > Amos > > > --- > Cyrus Home Page: http://asg.web.cmu.edu/cyrus > Cyrus Wiki/FAQ: http://cyruswiki.andrew.cmu.edu > List Archives/Info: http://asg.web.cmu.edu/cyrus/mailing-list.html > --- Cyrus Home Page: http://asg.web.cmu.edu/cyrus Cyrus Wiki/FAQ: http://cyruswiki.andrew.cmu.edu List Archives/Info: http://asg.web.cmu.edu/cyrus/mailing-list.html
Performance Monitoring?
Performance Monitoring? - What are people using to monitor their Cyrus Imap Server performance? - things we would like to monitor: connect to imap port to banner response time imap login time mailbox selection time imap process time (maybe create a new folder, move a few messages, delete them and expung) - We would also prefer graphing / trending, like a MRTG add-on. Thanks, Bill William B. Earle Computing & Information Technology University at Buffalo Voice: 716.645.6580 301 Computing Center Fax: 716.645.5972 Buffalo, NY 14260 --- Cyrus Home Page: http://asg.web.cmu.edu/cyrus Cyrus Wiki/FAQ: http://cyruswiki.andrew.cmu.edu List Archives/Info: http://asg.web.cmu.edu/cyrus/mailing-list.html
Re: Long startup time (1h 30m)
On Thu, 7 Oct 2004, Jules Agee wrote: > I've had this happen a few times, reconstructing the deliver db on > startup. I don't think this db is used for anything besides duplicate > delivery suppression, is it? Is there a recommended way to clean this > out periodically? Jules, Look at the man page for "cyr_expire" and "cyrus.conf". In recent releases you would have an entree similar to this in the EVENTS section of the cyrus.conf file: delprune cmd="cyr_expire -E 3" at=0400 - Bill > I rarely get duplicates that show up more than a > minute apart. I think it would be ideal to clean out all references to > messages more than a week old, since getting duplicates more than a week > apart would practically never happen. > > -Jules > > Dennis Sjogren wrote: > > Greetings! > > > > We've been running the Cyrus IMAP server for couple of months now. Apart > > from the server experiencing high load (we'll throw money at it and see > > what happens) from time to time, it's mostly been smooth sailing. > > > > However, one thing is a bit worrying. Startup of the Cyrus software > > takes about 1 hour and 30 minutes. For some reason, email services seems > > important to our users. :) Anyway, as any organization, we try to > > minimize downtime in core services as much as possible, but this long > > startup time really makes that difficult. > > > > I've googled ALOT, and found some users with similar problems or worse. > > Some guy had a startup time of 4 hours! However, no real solution has > > emerged. The wiki suggests checkpointing more often, but we already do > > that every 10 minutes. Shouldn't that be enough? > > > > Any suggestions, ideas, comments would be appreciated. Developers > > comments? Here are our specs: > > > > HOST: Sun Sunfire V240 > > CPU: 2 x 1.2GHz > > MEM: 2GB (will be upgraded to 4GB asap) > > DISK: (db) 2 x 10k RPM, internal, mirrored, UFS, nologging > > DISK: (spool) SAN (6120-system) > > > > Cyrus v2.2.5, BerkelyDB 4.2 + default patches > > Everything compiled with GCC 3.4.0 > > mailboxes.db about 10MB > > deliver.db about 18MB > > About 10k users on system, unknown number of mailboxes, though about 4M > > inodes are used on /var/spool/imap. > -- > Jules Agee > System Administrator > Pacific Coast Feather Co. > [EMAIL PROTECTED] x284 > --- > Cyrus Home Page: http://asg.web.cmu.edu/cyrus > Cyrus Wiki/FAQ: http://cyruswiki.andrew.cmu.edu > List Archives/Info: http://asg.web.cmu.edu/cyrus/mailing-list.html > --- Cyrus Home Page: http://asg.web.cmu.edu/cyrus Cyrus Wiki/FAQ: http://cyruswiki.andrew.cmu.edu List Archives/Info: http://asg.web.cmu.edu/cyrus/mailing-list.html
Re: CYRUS = GARBAGE
On Tue, 3 Feb 2004, Christian Schulte wrote: > Am Dienstag, 3. Februar 2004 10:50 schrieb Simon Matter: > > > Am Dienstag, 3. Februar 2004 01:53 schrieb Edward Rudd: > > >> That's not correct.. You can use mysql and have accounts automatically > > >> created.. > > >> If you install the autocreate patches.. > > >> http://email.uoa.gr/download/cyrus/cyrus-imapd-2.2.3/ > > > > > > I think you do not need to even patch 2.2.3 for this to work! > > > > For 'createonpost' and 'createonlogin', you need the patch AFAIK. I've > > included them in my rpms and they work pretty well. > > > > Simon > > I just took a look at the patches and I am wondering if they could replace the > autocreatequota: option which I am using right now. Do these patches support > creation of initial quota on user inboxes ? Just curious. Christian, I thought there used to be a "defaultquota" for imapd.conf, but all I could find was: >From imapd.conf(5) manpage: Cyrus Imap server version 2.2 autocreatequota: 0 If nonzero, normal users may create their own IMAP accounts by creating the mailbox INBOX. The user's quota is set to the value if it is positive, otherwise the user has unlimited quota. We don't use this feature here, YMMV. - Bill > -- > Christian >
Re: [POLL] Singleinstancestore
Jeremy, Sorry it took so long for me to respond. On Wed, 21 Jan 2004, Jeremy Rumpf wrote: [snip...] > From the TMS docs: Thanks! I asked out TSM folks here too, they came by with the manual. :) Slight difference, we use the backup / restore portion of TSM on our mail servers. It has the same paragraphs as the archive. :) [snip...] > So if you are using SSI, it seems that TSM will work the hard link magic for > you. And if it can't, it'll put down individual non-linked copies of the > files. This still could also lead to the situation where a selective restore > could end up utilizing more disk than the original files/links, but it looks > as if TSM tries harder than most to resolve things. Yes, it sounds like TSM will do the right thing. > Out of curiosity, how long does TSM take to run vs. the amount of mail data > you have? I've used TSM before, just not in this arena. We have a total of 12 spools on 4 different backend servers. The backups of the spools run concurrently, 3 per server, and TSM takes approx. 1 hour to backup the spools. Here is a sample report of 1 spool. Reporting imap.spool.07.log 01/22/04 23:02:38 01/22/04 23:02:38 Total number of objects inspected: 1,099,339 01/22/04 23:02:38 Total number of objects backed up: 25,000 01/22/04 23:02:38 Total number of objects updated: 0 01/22/04 23:02:38 Total number of objects rebound: 0 01/22/04 23:02:38 Total number of objects deleted: 0 01/22/04 23:02:38 Total number of objects expired: 15,234 01/22/04 23:02:38 Total number of objects failed: 8 01/22/04 23:02:38 Total number of bytes transferred: 1,014.87 MB 01/22/04 23:02:38 Data transfer time: 138.83 sec 01/22/04 23:02:38 Network data transfer rate:7,485.20 KB/sec 01/22/04 23:02:38 Aggregate data transfer rate:276.65 KB/sec 01/22/04 23:02:38 Objects compressed by:0% 01/22/04 23:02:38 Elapsed processing time: 01:02:36 01/22/04:23:02:38 Dump of /ssa/imap/spool/07 completed. - Bill
Re: [POLL] Singleinstancestore
Ken, We do NOT use singleinstancestore. We have several partitions for spools. Is the feature intelligent enough to determine which spools are on the same partition and hardlink those and create individual files for spools on separate partitions? - we use qmail and the deliver program, which is called one recpient, one message at a time, so this may not even effect us either way. We also would prefer to backup the files not a hardlink for restore purposes. I don't claim to know what the interaction with TSM and files vs. links are, but we are quite happy with files. TSM does an incremental forever backup, so only changes are backed up to tape. Does anyone know how the following scenario would be handled? scenario: message 1 is sent to studentB and facultyB - both are on the same partition TSM backs up message 1: - (file) for studentA - (link) for facultyB no changes have taken place to the file. studentA, no longer attends the university and the backup of their files expires. The tape copy is now gone. facultyB, accidentaly removes the file and requests a restore - does TSM have the file from studentA or just the link from facultyB? Thanks, - Bill On Wed, 21 Jan 2004, Ken Murchison wrote: > Alexey Melnikov wrote: > > > Ken Murchison wrote: > > > >> Does anyone out there actually disable singleinstancestore, and if so > >> why? > >> > >> Rob and I are working on some code changes and as part of them are > >> considering just having SIS always enabled. Is this going to create a > >> problem for any installations? > > > > > > If single instance store is implemented by checking Message-Id, this > > doesn't work in the real world. Outlook (or Outlook Express) keeps the > > original Message-Id when resending a message... > > We're not talking about duplicate suppression. We're talking about when > a message is sent to multiple recipients, we only have one instance of a > message in the spool and hardlinking to it from other mailboxes. > > -- > Kenneth Murchison Oceana Matrix Ltd. > Software Engineer 21 Princeton Place > 716-662-8973 x26 Orchard Park, NY 14127 > --PGP Public Key--http://www.oceana.com/~ken/ksm.pgp >
Re: Cyrus IMAPd 2.2.2-BETA Released
On Thu, 30 Oct 2003, Bill Earle wrote: > On Thu, 30 Oct 2003, Rob Siemborski wrote: > > > On Thu, 30 Oct 2003, Bill Earle wrote: > > > > > 1. It appears that the "--with-dbdir=/usr/local/BerkeleyDB.4.1" > > >does not get propogated in an include dir. > > >- the make fails in /ssa/Build/cyrus-imapd-2.2.2-BETA/imap > > > > Try --with-bdb-incdir and --with-bdb-libdir. > > Thanks Rob! As a temporary fix I had... > > export CPPFLAGS=-I/usr/local/BerkeleyDB.4.1/include > - then re-ran the configure, make depend and ran make just fine. I can successfully use CPPFLAGS and compile. I extracted a fresh copy and: - tried the following configure: ./configure \ --with-extraident=UB_2003_10_30 \ --with-auth=unix \ --with-cyrus-group=mail \ --prefix=/ssa/cyrus-2.2.2 \ --with-cyrus-prefix=/ssa/cyrus-2.2.2 \ --with-cyrus-user=cyrusadm \ --with-dbdir=/usr/local/BerkeleyDB.4.1 \ --with-bdb-incdir=/usr/local/BerkeleyDB.4.1/include \ --with-bdb-libdir=/usr/local/BerkeleyDB.4.1/lib \ --with-duplicate-db=skiplist \ --with-mboxlist-db=skiplist \ --with-seen-db=skiplist \ --with-subs-db=flat \ --with-tls-db=skiplist \ --with-idle=poll \ --with-openssl=/usr/local/ssl \ --with-perl=/usr/local/bin/perl \ --with-sasl=/usr/local \ --with-statedir=/var \ --disable-sieve \ --with-annotation-db=skiplist \ --with-pts-db=skiplist - make depend ; make - the build failed at the same place. ... gcc -c -I. -I.. -I. -I./../lib -I./../sieve -I./../et -I/usr/local/ssl/include -I/usr/local/include -DHAVE_CONFIG_H -g -O2 \ version.c version.c:46:16: db.h: No such file or directory make[1]: *** [version.o] Error 1 make[1]: Leaving directory `/ssa/Build/cyrus-imapd-2.2.2-BETA/imap' make: *** [all] Error 1 - as defined by --with-bdb-incdir=/usr/local/BerkeleyDB.4.1/include: bash-2.03# ls -la /usr/local/BerkeleyDB.4.1/include total 216 drwxr-sr-x 2 root staff512 Aug 22 14:36 . drwxr-sr-x 6 root staff512 Aug 22 14:36 .. -r--r--r-- 1 100 csfaculty1212 Jan 11 2002 cxx_common.h -r--r--r-- 1 100 csfaculty3967 Aug 7 2002 cxx_except.h -r--r--r-- 1 root root 75934 Aug 22 14:26 db.h -r--r--r-- 1 root root 25091 Aug 22 14:26 db_cxx.h
Re: Cyrus IMAPd 2.2.2-BETA Released
On Thu, 30 Oct 2003, Rob Siemborski wrote: > On Thu, 30 Oct 2003, Bill Earle wrote: > > > 1. It appears that the "--with-dbdir=/usr/local/BerkeleyDB.4.1" > >does not get propogated in an include dir. > >- the make fails in /ssa/Build/cyrus-imapd-2.2.2-BETA/imap > > Try --with-bdb-incdir and --with-bdb-libdir. Thanks Rob! As a temporary fix I had... export CPPFLAGS=-I/usr/local/BerkeleyDB.4.1/include - then re-ran the configure, make depend and ran make just fine. > Eventually we discovered that people were getting more creative at hiding > Berkely DB in some various location/name on their system, and we just > couldn't keep up. > > > 2. I have trouble building the Perl modules, even in previous > >releases, 2.1.15. > >- If I manually change the "prefix" for perl in "perl/Makefile" to: > > prefix = /usr/local > > - then the compile goes just fine > >- I use the configure option "--with-perl=/usr/local/bin/perl" > > do I need something else? > > What problem are you having? It may be more of a local problem. Going from memory here, I beleive what happened was the modules compile just fine, but perl scripts did not find them because they are installed in the "prefix" directory under /ssa/cyrus-, which is not part of our @INC path and can not be executed without some changes. Is there an option that would allow us to specify the perl prefix dir. separately? - Bill
Re: Cyrus IMAPd 2.2.2-BETA Released
I have a couple of questions for the build of cyrus-imapd-2.2.2-BETA. I get a couple of errors on building and was wondering if it was something I'm missing or maybe minor problem. - I am building on Solaris 8 / SPARC, gcc (GCC) 3.3 - configure options used: ./configure \ --with-extraident=UB_2003_10_30 \ --with-auth=unix \ --with-cyrus-group=mail \ --prefix=/ssa/cyrus-2.2.2 \ --with-cyrus-prefix=/ssa/cyrus-2.2.2 \ --with-cyrus-user=cyrusadm \ --with-dbdir=/usr/local/BerkeleyDB.4.1 \ --with-duplicate-db=skiplist \ --with-mboxlist-db=skiplist \ --with-seen-db=skiplist \ --with-subs-db=flat \ --with-tls-db=skiplist \ --with-idle=poll \ --with-openssl=/usr/local/ssl \ --with-perl=/usr/local/bin/perl \ --with-sasl=/usr/local \ --with-statedir=/var \ --disable-sieve 1. It appears that the "--with-dbdir=/usr/local/BerkeleyDB.4.1" does not get propogated in an include dir. - the make fails in /ssa/Build/cyrus-imapd-2.2.2-BETA/imap ... version.c version.c:46:16: db.h: No such file or directory make[1]: *** [version.o] Error 1 make[1]: Leaving directory `/ssa/Build/cyrus-imapd-2.2.2-BETA/imap' make: *** [all] Error 1 - db.h is in the --with-dbdir=/usr/local/BerkeleyDB.4.1 bash-2.03# find /usr/local/BerkeleyDB.4.1 -name db.h /usr/local/BerkeleyDB.4.1/include/db.h - the same problem continues for other modules looking for db.h 2. I have trouble building the Perl modules, even in previous releases, 2.1.15. - If I manually change the "prefix" for perl in "perl/Makefile" to: prefix = /usr/local - then the compile goes just fine - I use the configure option "--with-perl=/usr/local/bin/perl" do I need something else? - Bill On Tue, 28 Oct 2003, Rob Siemborski wrote: > I'm pleased to announce the release of Cyrus IMAPd 2.2.2-BETA. Currently, > this is planned to be the last "beta" release in the 2.2 series. It has a > number of new features and performance improvements including > much-expanded header caching and support for the IMAP SASL-IR and BINARY > extensions. This release also has a large number of bugfixes in the > Cyrus Murder support of virtdomain/altnamespace/unixhierachysep, as well > as a significant improvement in how Cyrus Master tracks processes. > > Full changes are available in doc/changes.html in the distribution, sites > that are upgrading to this version will also want to read > doc/install-upgrade.html. > > I'd personally like to thank Christian Schulte, Henrique de Moraes > Holschuh, David Carter, and Ken Murchison for their extensive > contributions to this release. Almost all of the fixes and new features > in this release are a direct result of their efforts. > > Download the release at: > ftp://ftp.andrew.cmu.edu/pub/cyrus/cyrus-imapd-2.2.2-BETA.tar.gz > or > http://ftp.andrew.cmu.edu/pub/cyrus/cyrus-imapd-2.2.2-BETA.tar.gz > > -Rob > > -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- > Rob Siemborski * Andrew Systems Group * Cyert Hall 207 * 412-268-7456 > Research Systems Programmer * /usr/contributed Gatekeeper > >
Re: Hashing database files?
On Fri, 24 Oct 2003, Rob Siemborski wrote: > On Fri, 24 Oct 2003, Bill Earle wrote: > > > - I hope this isn't the case. > > Why not? We are having contention problems with a single file and the multiple files may help. In our current config., deliverdb uses about 12 GB spread across our backends. This is version 2.0.16 and hashed deliver db's and we see no performance problems with 104 total files. (26 * 4) > > What configure option or patch is required to has the deliverdb's? > > Will hashing work with all 2 types of databse files, flat, DBD > > and skiplist? > > There isn't any. > > > Has anyone done this with mailboxes.db? > > This would probably be a really bad idea, given the way that mailboxes.db > is accessed. Is there a good document that describes how mailboxes.db and others are accessed? - Bill > > Would the configure option --enable-fulldirhash have any effect > > on the db's? > > No. > > -Rob > > -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- > Rob Siemborski * Andrew Systems Group * Cyert Hall 207 * 412-268-7456 > Research Systems Programmer * /usr/contributed Gatekeeper >
Hashing database files?
Hello Folks, I've looked over the docs, but can't locate a difinitive answer my questions. One item that sticks out, in the install-upgrade doc, under the heading Upgrading from 2.0.16 or earlier * The hashed deliver databases (used for duplicate delivery suppression and Sieve) have been merged into a single deliver.db database. You can safely remove the entire /var/imap/deliverdb directory structure after shutting down the server - I hope this isn't the case. What configure option or patch is required to has the deliverdb's? Will hashing work with all 2 types of databse files, flat, DBD and skiplist? Has anyone done this with mailboxes.db? Would the configure option --enable-fulldirhash have any effect on the db's? What I mean by hashing the deliverdb's is to split the db's by 1st initial of the usernameex: ls -1 deliverdb db/ deliver-a.db deliver-b.db deliver-c.db deliver-d.db deliver-e.db deliver-f.db deliver-g.db deliver-h.db deliver-i.db deliver-j.db deliver-k.db deliver-l.db deliver-m.db deliver-n.db deliver-o.db deliver-p.db deliver-q.db deliver-r.db deliver-s.db deliver-t.db deliver-u.db deliver-v.db deliver-w.db deliver-x.db deliver-y.db deliver-z.db Thanks in advance, Bill
Re: pine, horde and CRAM_MD5 not happy
On Mon, 30 Jun 2003, Ken Murchison wrote: > > Ezsra McDonald wrote: > > Since I was forced to migrate friday I have been > > unable to connect to the cyrus imap server with pine > > and horde. > > > > They are trying to use CRAM_MD5 authentication when I > > would rather they did not. > > > > How do I disable CRAM_MD5 as an authentication method > > or at least get it to work between cyrus and > > pine/horde? > > If you don't need them for any other applications, simply remove the > plugins from /user/lib/sasl2. If you only want to limit the mechs > offered by Cyrus, set the sasl_mech_list option in imapd.conf to a > space-separated list of mechs, eg: > > sasl_mech_list: PLAIN Ken's method certainly looks like the best option and would work globaly. One other option would be to change you pinerc file, look for: # List of SASL authenticators to disable. disable-these-authenticators= - Bill
Re: case insensitive delivery
Try calling a simple shell script and converting to lowercase before handing it off to deliver. ex. who=`echo $1|tr A-Z a-z` # convert username to lowercase - Bill On 28 Mar 2003, Chris Picton wrote: > Hi > > I am using postfix as a MTA, and it sends emails through with the case > unchanged to cyrus (either via cyrus, or deliver). > > Is there any way to get cyrus to accept for both chrisp and ChrisP, and > deliver them to the same mailbox chrisp? > > Regards > -- > -+-- > Chris Picton | PGP Key ID: 9D28A988 (wwwkeys.pgp.net) > Solutions Developer | PGP Key Fingerprint: > Tangent Systems | 2B46 29EA D530 79EC D9EA 3ED0 229D 6DD6 9D28 A988 > [EMAIL PROTECTED] | http://www.tangent.co.za/keys/chrisp.asc > -+-- >
Re: Bare newlines
On Fri, 28 Mar 2003, John Alton Tamplin wrote: > Sebastian Konstanty Zdrojewski wrote: > > > I was attempting to move an email message from a local account onto an > > IMAP directory located on Cyrus. The client I am using is Netscape > > Mail and usually it gives me no problems. This time, I am not able to > > move the message because the client gives the following error: > > > > Unable to perform operation. Message contains bare newlines. > > > > The message has an attached PDF file (2Mb size) and a signature with > > image integrated. Any idea of what can be my problem? > > In IMAP, messages are required to have \r\n as line terminators, not > just \n. Many other IMAP servers (or mail clients' local folders) do > not enforce strict RFC compliance, while Cyrus does -- when we converted > from UW-IMAP to Cyrus I had to write code to munge messages into the > proper format (8 bit or null data in headers, and bogus headers are the > other two issues you will likely run into). John, How are you dealing with "8 bit or null data in headers, and bogus headers"? - Bill > > -- > John A. Tamplin Unix System Administrator > Emory University, School of Public Health +1 404/727-9931 > > >