[rt-users] Wide character in subroutine entry at /usr/share/perl5/vendor_perl/Cache/FileBackend.pm line 201
Recently when trying to edit some queues configuration in rt 4.0.4 and I got: Wide character in subroutine entry at /usr/share/perl5/vendor_perl/Cache/FileBackend.pm line 201. where that's subroutine _Build_Unique_Key and the line is return sha1_hex( $p_key ); The workarounds I came up was doing: use Encode qw(encode_utf8); return sha1_hex( encode_utf8($p_key) ); or simply disabling GPG. What's correct fix for that? -- Arkadiusz MiśkiewiczPLD/Linux Team arekm / maven.plhttp://ftp.pld-linux.org/ RT Training Sessions (http://bestpractical.com/services/training.html) * Barcelona, Spain November 28 & 29, 2011
[rt-users] attachment filename corrupted
Filename attachments sent with headers like below are getting corrupted name in rt (4.0.6 here). For example instead of "rozmowy wstępne - rozliczenie za kwiecień.xlsx" rt displays truncated name "Download rozmowy wstępne -; application/vnd.openxmlformats-officedocument.spreadsheetml.sheet 6.2k" Any chances to get this fixed? Content-Type: application/vnd.openxmlformats-officedocument.spreadsheetml.sheet; name="rozmowy =?UTF-8?B?d3N0xJlwbmUgLSByb3psaWN6ZW5pZSB6YSBrd2llY2llxYQueGxzeA==?=" Content-Transfer-Encoding: base64 Content-Disposition: attachment; filename*0*=UTF-8''%72%6F%7A%6D%6F%77%79%20%77%73%74%C4%99%70%6E%65%20%2D; filename*1*=%20%72%6F%7A%6C%69%63%7A%65%6E%69%65%20%7A%61%20%6B%77%69%65; filename*2*=%63%69%65%C5%84%2E%78%6C%73%78 -- Arkadiusz Miśkiewicz, arekm / maven.pl
Re: [rt-users] attachment filename corrupted
On Wednesday 27 of June 2012, Ruslan Zakirov wrote: > Hi, > > Should work in 4.0.6. Ohh, seems to be working for *new* emails. Thanks! -- Arkadiusz Miśkiewicz, arekm / maven.pl
[rt-users] rt 4.0.8 and unknown charsets
Hi, rt 4.0.8 installation and rt-mailgate script quite often cannot deliver message to rt because: [pid 8672] write(2, "RT server error.\n\nThe RT server which handled your email did not behave as expected. It\nsaid:\n\nUnknown encoding 'iso-8859-8-i' at /usr/share/perl5/vendor_perl/RT/I18N.pm line 542\n\nStack:\n [/usr/share/perl5/5.12.4/Carp.pm:62]\n [/usr/lib/perl5/vendor_perl/5.12.0/i686-pld-linux-thread- multi/Encode.pm:189]\n [/usr/share/perl5/vendor_perl/RT/I18N.pm:542]\n [/usr/share/perl5/vendor_perl/RT/I18N.pm:214]\n [/usr/share/perl5/vendor_perl/RT/I18N.pm:210]\n [/usr/share/perl5/vendor_perl/RT/EmailParser.pm:282]\n [/usr/share/perl5/vendor_perl/RT/Interface/Email.pm:1433]\n [/usr/share/rt/html/REST/1.0/NoAuth/mail-gateway:61]\n\n", 621) = 621 iso-8859-8-i is actually valid encoding (http://en.wikipedia.org/wiki/ISO-8859-8-I) The question is - can rt be configured/changed that way it treats unknown charsets as ascii or utf8? I really prefer to have borked charset than to loose message. -- Arkadiusz Miśkiewicz, arekm / maven.pl We're hiring! http://bestpractical.com/jobs
Re: [rt-users] rt 4.0.8 and unknown charsets
On Friday 09 of November 2012, Thomas Sibley wrote:
> On 11/09/2012 01:47 AM, Arkadiusz Miśkiewicz wrote:
> > iso-8859-8-i is actually valid encoding
> > (http://en.wikipedia.org/wiki/ISO-8859-8-I)
>
> Try putting this in your RT_SiteConfig.pm:
>
> use Encode;
> use Encode::Alias;
> define_alias("iso-8859-8-i" => "iso-8859-8");
>
> and then restarting RT.
That worked for this case, thanks!
Now I'm interested in more generic solution, so I won't have to play manually
with every (random) charset someone is putting :)
--
Arkadiusz Miśkiewicz, arekm / maven.pl
We're hiring! http://bestpractical.com/jobs
[rt-users] adding new system RT System's saved searches
Hi, I can edit system searches by settings->saved searches->"unowned tickets" (for example) -> "You can also edit the predefined search itself: Search - Unowned Tickets" Now I want add new system search like "owned tickets" for example. When editing existing system unowned ticket search the description for it is "[_1] newest unowned tickets". What is "[_1]" meaning and what to use as description there if I want new _system_ search? rt 4.0.8 Thanks, -- Arkadiusz Miśkiewicz, arekm / maven.pl We're hiring! http://bestpractical.com/jobs
Re: [rt-users] adding new system RT System's saved searches
On Wednesday 14 of November 2012, Kevin Falcone wrote: > On Mon, Nov 12, 2012 at 11:14:25AM +0100, Arkadiusz Miśkiewicz wrote: > > I can edit system searches by settings->saved searches->"unowned tickets" > > (for example) -> "You can also edit the predefined search itself: Search > > - Unowned Tickets" > > > > Now I want add new system search like "owned tickets" for example. > > Unfortunately, you can't do this from the UI. > You can only do it with an initialdata file (see the etc/initialdata > file we ship with RT and the docs/initialdata.pod available in git. > > The best you can do from the UI is to make a group saved search for a > group everyone is in. This is a limitation we hope to remove in a > future major release of RT. Hm, I did what I described earlier (editing "You can also edit the predefined search itself: Search - Unowned Tickets"), modified search rules to match my needs and then saved using "Save as New". It got saved in "RT System's saved searches", so looks like adding new system search is possible from UI right now. Thanks, -- Arkadiusz Miśkiewicz, arekm / maven.pl We're hiring! http://bestpractical.com/jobs
Re: [rt-users] adding new system RT System's saved searches
On Wednesday 14 of November 2012, Thomas Sibley wrote: > On 11/14/2012 08:42 AM, Arkadiusz Miśkiewicz wrote: > > It got saved in "RT System's saved searches", so looks like adding new > > system search is possible from UI right now. > > Yes, but it's unfortunately subtly different than the stock searches, > and it won't be visible to all users (just superusers). Such search added to default "rt at glance" works also for regular users and that was my goal. Thanks, -- Arkadiusz Miśkiewicz, arekm / maven.pl We're hiring! http://bestpractical.com/jobs
[rt-users] "My reminders" showing differently than for example linked tickets
In rt when two tickets are linked and one gets closed then the other one shows ticket title crossed with line like: "Refers to : (Create) ---954173: abc something---" (title is crossed with line). I wonder if "My reminders" could do the same? Right now I don't immediately see which reminders are for closed tickets and which are for open ones. Unification here would be nice IMO (so treat this as feature requst, too). -- Arkadiusz Miśkiewicz, arekm / maven.pl -- RT training in Amsterdam, March 20-21: http://bestpractical.com/services/training.html Help improve RT by taking our user survey: https://www.surveymonkey.com/s/N23JW9T
[rt-users] [PATCH REQ for 4.2.2] callback for Modify.html
Hi. Update.html contains nice callback for various tasks (like checking form fields etc): $m->callback( CallbackName => 'BeforeUpdate', ARGSRef => \%ARGS, skip_update => \$skip_update, ... but Modify.html doesn't have such thing. Was there any reason for not adding BeforeModify callback in Modify.html? It's logical to have the same sets of callbacks for similar actions IMO. This patch below works (hand made, line wrapped, so likely won't apply directly). Please add such modification to 4.2.2. Thanks. --- html/Ticket/Modify.html~ 2013-11-25 10:25:50.0 +0100 +++ html/Ticket/Modify.html2013-11-26 13:36:25.847864830 +0100 @@ -77,6 +77,9 @@ my @results; my $skip_update = 0; +$m->callback( CallbackName => 'BeforeModify', ARGSRef => \%ARGS, skip_update => \$skip_update, + results => \@results, TicketObj => $TicketObj ); + # Now let callbacks have a chance at editing %ARGS $m->callback( TicketObj => $TicketObj, CustomFields => $CustomFields, ARGSRef => \%ARGS, skip_update => \$skip_update ); -- Arkadiusz Miśkiewicz, arekm / maven.pl
Re: [rt-users] [PATCH REQ for 4.2.2] callback for Modify.html
On Tuesday 26 of November 2013, Kevin Falcone wrote: > On Tue, Nov 26, 2013 at 01:58:47PM +0100, Arkadiusz Miśkiewicz wrote: > > Was there any reason for not adding BeforeModify callback in Modify.html? > > It's logical to have the same sets of callbacks for similar actions IMO. > > > > This patch below works (hand made, line wrapped, so likely won't apply > > directly). Please add such modification to 4.2.2. Thanks. > > Your patch adds a callback right before another callback which passes > more useful information and can be used to accomplish the same goals. > The existing callback is even visible in your patch. > > What does your new callback do that the Default callback doesn't? That's a good question + #rt irc conversation explained few things for me. It's not needed at all. "Default" will be enough. -- Arkadiusz Miśkiewicz, arekm / maven.pl
Re: [rt-users] [PATCH REQ for 4.2.2] callback for Modify.html
On Tuesday 26 of November 2013, Arkadiusz Miśkiewicz wrote: > On Tuesday 26 of November 2013, Kevin Falcone wrote: > > On Tue, Nov 26, 2013 at 01:58:47PM +0100, Arkadiusz Miśkiewicz wrote: > > > Was there any reason for not adding BeforeModify callback in > > > Modify.html? It's logical to have the same sets of callbacks for > > > similar actions IMO. > > > > > > This patch below works (hand made, line wrapped, so likely won't apply > > > directly). Please add such modification to 4.2.2. Thanks. > > > > Your patch adds a callback right before another callback which passes > > more useful information and can be used to accomplish the same goals. > > The existing callback is even visible in your patch. > > > > What does your new callback do that the Default callback doesn't? > > That's a good question + #rt irc conversation explained few things for me. > It's not needed at all. "Default" will be enough. One thing is missing in Default though. results array. # Now let callbacks have a chance at editing %ARGS $m->callback( TicketObj => $TicketObj, CustomFields => $CustomFields, ARGSRef => \%ARGS, skip_update => \$skip_update ); While new BeforeModify callback allows us to change error messages for example. +$m->callback( CallbackName => 'BeforeModify', ARGSRef => \%ARGS, skip_update => \$skip_update, + results => \@results, TicketObj => $TicketObj ); Could Default callback call be changed to allow results array modifications? Or is there any other way to change results from Default callback? -- Arkadiusz Miśkiewicz, arekm / maven.pl
Re: [rt-users] REST mail-gateway using 100% cpu
On Wednesday 15 of January 2014, Kevin Falcone wrote: > It will be re-queued and tried again > when it fails to submit to RT, so it'll be somewhere in your queues I find it to not always be true. For example when our mysql died then rt- mailgate didn't return any error BUT rt itself created bounces with "Ticket creation failed: Mail delivery failed: returning message to sender". So this means customers messages were lost (we fortunately recovered these from other sources), not requeued obviously and customers were scared with "Ticket could not be created due to an internal error" mail bodies. I would even mark this as a "important bug" (as loosing messages is important) but well... rt 4.0.18 > -kevin -- Arkadiusz Miśkiewicz, arekm / maven.pl
[rt-users] rt being silent about not saving attachments in case of sql query failure
Hello. I'm using rt 4.2.7 and sending mails with attachments to it. Unfortunately due to mysql db problems only core mail message is being saved without any attachments: [9761] [Mon Sep 29 11:19:03 2014] [critical]: Attachment insert failed: Row size too large (> 8126). Changing some columns to TEXT or BLOB or using ROW_FORMAT=DYNAMIC or ROW_FORMAT=COMPRESSED may help. In current row format, BLOB prefix of 768 bytes is stored inline. (/usr/share/perl5/vendor_perl/RT/Attachment.pm:219) It would make sense to store any information about this failure inside of rt ticket (something like "failed to save attachments") instead of silently (well, there is this log) dropping it. Thanks, -- Arkadiusz Miśkiewicz, arekm / maven.pl -- RT Training November 4 & 5 Los Angeles http://bestpractical.com/training
[rt-users] sphinx weirdness
lled' ) ); ++-++++-+-+-+--+---+ | id | select_type | table | type | possible_keys | key | key_len | ref | rows | Extra | ++-++++-+-+-+--+---+ | 1 | SIMPLE | AttachmentsIndex_3 | ref| query | query | 767 | const |3 | Using where with pushed condition | | 1 | SIMPLE | Attachments_2 | eq_ref | PRIMARY,Attachments2 | PRIMARY | 4 | rt3.AttachmentsIndex_3.id |1 | Using where | | 1 | SIMPLE | Transactions_1 | eq_ref | PRIMARY,Transactions1 | PRIMARY | 4 | rt3.Attachments_2.TransactionId |1 | Using where | | 1 | SIMPLE | main | eq_ref | PRIMARY,tickets_status | PRIMARY | 4 | rt3.Transactions_1.ObjectId |1 | Using where | ++-----+----+++-+-+-+--+---+ 4 rows in set (0.00 sec) Any idea what is going on? Additional subject searching shouldn't cause such problem since it's ORed with the rest of that part of query. -- Arkadiusz Miśkiewicz, arekm / ( maven.pl | pld-linux.org )
Re: [rt-users] sphinx weirdness
On Wednesday 25 of February 2015, Arkadiusz Miśkiewicz wrote: > mysql> EXPLAIN SELECT COUNT(DISTINCT main.id) FROM Tickets main JOIN > Transactions Transactions_1 ON ( Transactions_1.ObjectType = 'RT::Ticket' ) > AND ( Transactions_1.ObjectId = main.id ) LEFT JOIN Attachments > Attachments_2 ON ( Attachments_2.TransactionId = Transactions_1.id ) LEFT > JOIN AttachmentsIndex AttachmentsIndex_3 ON ( AttachmentsIndex_3.id = > Attachments_2.id ) WHERE (main.IsMerged IS NULL) AND (main.Status != > 'deleted') AND (main.Type = 'ticket') AND ( ( ( main.Subject LIKE > '%inconventus%' OR ( AttachmentsIndex_3.query = 'inconventus' ) ) ) AND > ( main.Status = 'new' OR main.Status = 'open' OR main.Status = 'stalled' ) ); > ++-++---+++-+---+--++ > | id | select_type | table | type | possible_keys | > key| key_len | ref | rows | Extra >| > ++-++---+++-+---+--++ > | 1 | SIMPLE | main | range | PRIMARY,tickets_status | > tickets_status | 195 | NULL | 2228 | Using index > condition; Using where | > | 1 | SIMPLE | Transactions_1 | ref | Transactions1 | > Transactions1 | 70 | const,rt3.main.id |1 | Using where; Using > index | > | 1 | SIMPLE | Attachments_2 | ref | Attachments2 | > Attachments2 | 4 | rt3.Transactions_1.id |4 | Using index >| > | 1 | SIMPLE | AttachmentsIndex_3 | ALL | NULL | > NULL | NULL| NULL | 20 | Using where; Using > join buffer (Block Nested Loop) | > ++-++---+++-+---+--++ > 4 rows in set (0.00 sec) Ok, mysql is too smart! For sphinx to work mysql needs to first query AttachmentsIndex_3 and then make joins to it. Otherwise sphinx won't work. That's due to sphinx architecture. Here mysql is too smart and differently optimizes query thus breaking sphinx support in rt. Now I've tried FORCE INDEX and such but wasn't able to force mysql to first query AttachmentsIndex_3. There is STRAIGHT_JOIN that forces joins orders, so maybe that is some solution. Other ideas? -- Arkadiusz Miśkiewicz, arekm / ( maven.pl | pld-linux.org )
Re: [rt-users] sphinx weirdness [explanation and possible solution]
On Wednesday 25 of February 2015, Arkadiusz Miśkiewicz wrote: > Ok, mysql is too smart! For sphinx to work mysql needs to first query > AttachmentsIndex_3 and then make joins to it. Otherwise sphinx won't work. > That's due to sphinx architecture. > > Here mysql is too smart and differently optimizes query thus breaking > sphinx support in rt. > > Now I've tried FORCE INDEX and such but wasn't able to force mysql to first > query AttachmentsIndex_3. > > There is STRAIGHT_JOIN that forces joins orders, so maybe that is some > solution. STRAIGHT_JOIN also won't work since mysql still is able to make changes and optimizations to the query. Fortunately code below seems to be working - using UNION and separate, simple sphinx subquery: SELECT DISTINCT main.id FROM Tickets main JOIN Transactions Transactions_1 ON ( Transactions_1.ObjectType = 'RT::Ticket' ) AND ( Transactions_1.ObjectId = main.id ) JOIN Attachments Attachments_2 ON ( Attachments_2.TransactionId = Transactions_1.id ) JOIN ((SELECT ai.id FROM AttachmentsIndex ai WHERE ai.query = 'inconventus') UNION (SELECT ti.id FROM Tickets ti WHERE ti.Subject LIKE '%inconventus%')) u ON u.id=Attachments_2.id WHERE (main.IsMerged IS NULL) AND (main.Status != 'deleted') AND (main.Type = 'ticket') AND ( main.Status = 'new' OR main.Status = 'open' OR main.Status = 'stalled' ); Devs, could you please change querying code, so that sphinx will always get its own subquery? That will always work because mysql query optimizer will not be able to mess with sphinx query. Optimizer will be able only to optimize combining sphinx results with the rest of query though and that's ok and desired. Sphinx subquery will always be simple, like SELECT ai.id FROM AttachmentsIndex ai WHERE ai.query = 'inconventus' Thanks, -- Arkadiusz Miśkiewicz, arekm / ( maven.pl | pld-linux.org )
[rt-users] forwarding correspondence lacks important data in headers
Hi. rt 4.2.12, scenario - there is a ticket, user enters new correspondence via web interface which is send via email to requestor. Then user clicks "forward" near that correspondence and forwards it to third email address / third person. Unfortunately that email sent to third person is lacking important information. It lacks From, To, Subject headers in attached forwarded message. Example - this is what third person gets as attached forwarded message, no more headers than these: MIME-Version: 1.0 X-RT-Interface: Web X-Mailer: MIME-tools 5.505 (Entity 5.505) X-RT-Original-Encoding: utf-8 Content-Type: multipart/mixed; boundary="--=_14261232133-20661-12746" Message-ID: In-Reply-To: References: RT-Send-CC: Content-Length: 0 X-UID: 0 [here body of forwarded message follows] Is this how this always worked? If not, there is a bug somewhere. I would expect to see proper From, To, Subject, Date (like in original correspondence) and possibly other headers, too. Also there is ncorrect Content-Length in it. -- Arkadiusz Miśkiewicz, arekm / ( maven.pl | pld-linux.org )
Re: [rt-users] forwarding correspondence lacks important data in headers
On Tuesday 08 of September 2015, Alex Vandiver wrote: > On Tue, Sep 08, 2015 at 08:56:53AM +0200, Arkadiusz Miśkiewicz wrote: > > Hi. > > > > rt 4.2.12, scenario - there is a ticket, user enters new correspondence > > via web interface which is send via email to requestor. > > > > Then user clicks "forward" near that correspondence and forwards it to > > third email address / third person. > > > > Unfortunately that email sent to third person is lacking important > > information. It lacks From, To, Subject headers in attached forwarded > > message. > > This should have been fixed in 4.2.10: > https://bestpractical.com/release-notes/rt/4.2.10 > > https://github.com/bestpractical/rt/commit/134a478be1b63ca418380fc355923ea > 849812bfa > > Are you still seeing a lack of such headers for tickets created since > your upgrade to 4.2.12? Tickets created _prior_ to an upgrade to > 4.2.10 will still lack the headers in their Create transactions, of > course. I've checked forwarding of a single correspondence (one correspondence entry, not entire ticket) where ticket and that correspondence was created on 4.2.12. Unfortunately the problem persist - headers are missing information. How to reproducte: - create ticket via web UI - add correspondence with some text and attach pdf file to it; via web UI) - click forward near that correspondence (we want to forward only that one correspondence entry, not entire ticket); again via web UI - enter some email address "email address" receives message with missing headers. > > I would expect to see proper From, To, Subject, Date (like in original > > correspondence) and possibly other headers, too. Also there is incorrect > > Content-Length in it. > > Hm -- the content-length problem doesn't seem to be replicated on > issues.bestpractical.com. Are you seeing it with a vanilla 4.2.12 > install? > - Alex -- Arkadiusz Miśkiewicz, arekm / ( maven.pl | pld-linux.org )
Re: [rt-users] What's huge RT/Mysql
On Wednesday 04 of November 2015, Albert Shih wrote: > Hi everyone, > > I would like to know without any customisation how can a RT instance with > mysql backend can goes. Almost 2 000 000 tickets here (total), mysql DB is about 120GB, two rotational SATA disks in raid1 array. > As I understand it's the number of tickets who can be a issue. For normal usage it works fast. The number of tickets is only a problem for searching. Here we still use sphinx and searching is quite slow. The other problem is that any action like searching blocks rt user session totally. So single user can't, for example, view other tickets in second browser tab while his search is in progress. That's major pain if you do a lot of searching. Going to try mysql builtin fulltext next (that won't fix session problem but has potential to be faster). Also for big mysql DB you need to plan backup well. > And same question with specific hardware (like mysql on SSD). IMO SSD is not required. Obviously will be much faster than rotational disks but our instance works well without ssd disks. > > Regards. > > JAS > -- > Albert SHIH -- Arkadiusz Miśkiewicz, arekm / ( maven.pl | pld-linux.org )
Re: [rt-users] What's huge RT/Mysql
On Wednesday 11 of November 2015, Daniel Schwager wrote: > Hi JAS, > > > For normal usage it works fast. The number of tickets is only a problem > > for searching. Here we still use sphinx and searching is quite slow. > > .. > > > Going to try mysql builtin fulltext next (that won't fix session problem > > but has potential to be faster). > > Currently, we are going to setup a system with sphinx & mysql 5.6 because > of the sphinx-fulltext search feature. > > So, do you think mysql-5.7 with enabled mysql-fulltext feature is faster? Why 5.7? For 5.7 is to early to use in production IMO. I plan to go with 5.6 + fulltext. I didn't do any measures yet but it should be on par with sphinx I think. The main advantage of mysql fulltext is that there is no gap between ticket coming into system and sphinx indexer being run. > Can RT use the builtin mysql fulltext search by default without any > changes? It can in recent versions. sphinx is considered deprecated I think (and buggy; doesn't make proper sql queries because mysql optimization engine is allowed to optimize these in such way that breaks sphinx way of working). > Best regards > Danny -- Arkadiusz Miśkiewicz, arekm / ( maven.pl | pld-linux.org )
