Daily Log for #alfresco IRC Channel

Alfresco discussion and collaboration. Stick around a few hours after asking a question.

Official support for Enterprise subscribers: support.alfresco.com.

Joining the Channel:

Join in the conversation by getting an IRC client and connecting to #alfresco at Freenode. Our you can use the IRC web chat.

More information about the channel is in the wiki.

Getting Help

More help is available in this list of resources.

Daily Log for #alfresco

2019-12-20 08:35:47 GMT <alfresco-discord> <IanW> We're also getting fairly close to deciding to migrate away - my boss is very keen on google docs via gsuite not least because we already seem to have a lot of stuff going on via personal google docs accounts instead of using Alfresco - it's now pretty rare for anybody to use Share for vanilla docs/projects - basically if I'm not involved it doesn't happen...

2019-12-20 08:46:07 GMT <alfresco-discord> <yreg> I wonder if Alfresco is being aware that all they have been doing the last few years is ||punching their own face|| ...

2019-12-20 08:58:50 GMT <alfresco-discord> <IanW> It does feel a bit like that doesn't it

2019-12-20 10:05:34 GMT <alfresco-discord> <monica> Hello everyone. I need a help in preview issue. I am using alfresco 5.2 and not able to preview only pdf documents on quickshare page. Only pdf types of documents are not previewing on quickshare/external share page. all other mimetypes are working fine. Any ideas guys ?

2019-12-20 10:06:17 GMT <alfresco-discord> <monica> I have checked for the dependencies. seems fine

2019-12-20 10:06:59 GMT <alfresco-discord> <monica> Please guide.

2019-12-20 10:29:58 GMT <alfresco-discord> <monica> getting ""could not load PDF content" error on page.

2019-12-20 10:36:09 GMT <angelborroy> No stacktrace in catalina.out?

2019-12-20 10:53:40 GMT <alfresco-discord> <monica> nope.. thats the issue.. not printing anytthing on console or catalina.out

2019-12-20 11:54:18 GMT <angelborroy> AFaust I’m trying to improve tracking performance

2019-12-20 11:54:35 GMT <AFaust> heya...

2019-12-20 11:54:38 GMT <angelborroy> Wondering if GZIP request could help

2019-12-20 11:54:50 GMT <angelborroy> Any thoughts on that?

2019-12-20 11:56:52 GMT <angelborroy> Well, in fact GZIP applied to responses, I mean

2019-12-20 11:57:29 GMT <AFaust> I don't think that network / HTTP throughput really puts much of a dent in the performance, but it would be one of the low-hanging fruits, especially when content is transferred for full text indexing

2019-12-20 11:57:51 GMT <angelborroy> Ok, I’ll try that

2019-12-20 11:58:08 GMT <angelborroy> I’ve also replaced delete by a search-not-exist-delete sequence

2019-12-20 11:58:25 GMT <angelborroy> But it doesn’t seem to helps much in terms of performance

2019-12-20 11:58:47 GMT <angelborroy> And also applied a logic for DB_ID_RANGE to skip useless transactions

2019-12-20 12:02:49 GMT <AFaust> Ok - I already started writing a response that those changes are more important - so good to hear they have already been covered

2019-12-20 12:03:33 GMT <angelborroy> I will deliver a RC to Performance Team next week

2019-12-20 12:03:38 GMT <angelborroy> Let’s see if that helps

2019-12-20 12:03:49 GMT <AFaust> Though for the search-not-exist-delete sequence my suggestion would be to support a "index up to txn XY in bulk mode" option, completely disabling that "delete" handling

2019-12-20 12:04:50 GMT <angelborroy> Yep, that would be an interesting alternative

2019-12-20 12:05:11 GMT <angelborroy> I’ll take a look also at this

2019-12-20 12:05:37 GMT <AFaust> E.g. when I reindex ~5 million txns while the system is active, I should be able to run in this mode until txn 4950000, and then switch to index the remainder, which could contain changes to nodes I just bulk-indexed

2019-12-20 12:06:21 GMT <angelborroy> You mean a manual setting?

2019-12-20 12:06:27 GMT <angelborroy> That seems dangerous

2019-12-20 12:06:54 GMT <AFaust> Well, how else would you allow for a proper "this txn is safe - the next one isn't" distinction?

2019-12-20 12:07:36 GMT <angelborroy> Yep, that it’s (almost) impossible

2019-12-20 12:08:13 GMT <AFaust> The search-not-exist-delete sequence has a valid requirement to exist - and it is only the sysadmin who is able to determine when that requirement is not relevant, e.g. when doing a bulk re-index (e.g. after an upgrade)

2019-12-20 12:08:28 GMT <AFaust> (or a re-shard)

2019-12-20 12:10:06 GMT <AFaust> If I remember correctly from my 100m tests, that delete handling can easily consume 40-80% of time for a massive re-index.

2019-12-20 12:10:22 GMT <angelborroy> Good to know

2019-12-20 12:11:49 GMT <AFaust> Skipping useless txns in 1h intervals is the next best "time-killer". If that is already improved for DB_ID_RANGE, then that is fine. But it should also be improved for more general situations.

2019-12-20 12:12:09 GMT <AFaust> E.g. I recently had a customer who used Alfresco since late 1.x / early 2.x times...

2019-12-20 12:13:01 GMT <AFaust> Due to ancient, crappy Alfresco patches / schema upgrades, their alf_transaction table contained commit_time_ms values like "3" (yes, 3 milliseconds after Unix epoch start)

2019-12-20 12:13:55 GMT <AFaust> So tracking with 1h intervals would have taken 8-9 days just to catch up to 2019 due to the "select 1h window" increment overhead

2019-12-20 12:14:38 GMT <angelborroy> Another interesting finding

2019-12-20 12:15:03 GMT <AFaust> Every customer should have frequent periods with little activity, where there might not even be one transaction, e.g. weekends, when little to no activity is happening, unless there are some CRONs in other systems which generate documents for Alfresco

2019-12-20 12:15:24 GMT <AFaust> During re-indexing, those periods also waste a lot of time by skipping in 1h increments.

2019-12-20 12:16:43 GMT <AFaust> Improvement #1 would be to have that query window configurable - improvement #2 would be to refactor the query to no longer require a fixed window of time (e.g. sort by commit_time_ms + FETCH FIRST XX ROWS)

2019-12-20 12:16:47 GMT <angelborroy> Probably we can have something like 1 hour or 2000 transactions when getting the list

2019-12-20 12:17:23 GMT <AFaust> Well, those are currently the limits, but whatever hits first ,so max 1 hour AND max 2000 transactions

2019-12-20 12:17:34 GMT <AFaust> not either-or

2019-12-20 12:18:00 GMT <angelborroy> Yes, your English is better than mine :D

2019-12-20 12:18:15 GMT <AFaust> Well, in this case it is not, because "either-or" is not correct either...

2019-12-20 12:19:00 GMT <angelborroy> I’m playing these days with YourKit

2019-12-20 12:19:02 GMT <AFaust> because the current state is actually "either-or" when AND-combining the limits

2019-12-20 12:19:05 GMT <angelborroy> Nice tool to play with

2019-12-20 12:22:15 GMT <AFaust> Sure... it simplifies analysis, which otherwise would be a bit harder to do with just the default Java tools (but is doable without requiring such commercial tools)

2019-12-20 12:22:59 GMT <angelborroy> We have a license in Alfreso for YourKit, this is why I’m using it

2019-12-20 12:28:04 GMT <AFaust> As a software vendor, I would expect Alfresco to have license for such tools to safe on analysis effort

2019-12-20 15:02:27 GMT <alfresco-discord> <yreg> Good News everyone, Alfresco has just publicly announced DevCon !

2019-12-20 15:02:37 GMT <alfresco-discord> <yreg> @here

2019-12-20 15:03:04 GMT <alfresco-discord> <MartinM> 🥳

2019-12-20 15:04:56 GMT <alfresco-discord> <Hans> London?? 😦

2019-12-20 15:05:42 GMT <alfresco-discord> <yreg> Yep, I just got the mail and it says : > Where: London, UK (exact venue location will be revealed in the new year)

2019-12-20 15:07:30 GMT <AFaust> Are they sending out emails manually or are you on some "preferred audience" list?

2019-12-20 15:10:03 GMT <alfresco-discord> <yreg> I guess being a partner has its perks 😄

2019-12-20 15:10:21 GMT <alfresco-discord> <mbui> I've heard rumours that the dates will be 5th to 7th of May :>

2019-12-20 15:10:26 GMT <alfresco-discord> <yreg> I only started guetting those technical announcement after I signed up for a new partners portal I guess

2019-12-20 15:32:46 GMT <alfresco-discord> <dgradecak> didn't they say it will be a sunny place 😉

2019-12-20 15:43:15 GMT <angelborroy> This is the first news from my side

2019-12-20 15:43:17 GMT <angelborroy> :D

2019-12-20 15:51:50 GMT <alfresco-discord> <digcat> and still nothing on https://devcon.alfresco.com/

2019-12-20 16:42:06 GMT <alfresco-discord> <Francesco Corti> Thank you for the patience guys

2019-12-20 16:42:23 GMT <alfresco-discord> <Francesco Corti> The plan is announce the DevCon before Christmas

2019-12-20 16:43:06 GMT <alfresco-discord> <Francesco Corti> I want to believe the Marketing team is announcing the DevCon also in the open, not only in the partner portal

2019-12-20 16:43:44 GMT <alfresco-discord> <Francesco Corti> Until that time, partners who are also community members can give a hand and share the info

2019-12-20 16:44:34 GMT <alfresco-discord> <Francesco Corti> It is still the same community event (in the broad term: enthusiasts in general).

End of Daily Log

The other logs are at http://esplins.org/hash_alfresco