Gajim - 2017-10-03

  1. andrey.g Now it's much better with logs, thanks.
  2. andrey.g 10/02/2017 23:55:16 (I) nbxmpp.client_nb raising event from transport: :::::DATA RECEIVED:::: _____________ <a h='35' xmlns='urn:xmpp:sm:2'/> _____________ 10/02/2017 23:55:16 (D) nbxmpp.smacks Got ack for outgoing stanzas (our h: 63, server h: 35), removing 1 messages from queue... 10/02/2017 23:55:16 (I) nbxmpp.transports_nb pollin called, state == CONNECTED 10/02/2017 23:55:16 (I) nbxmpp.client_nb raising event from transport: :::::DATA RECEIVED::::
  3. andrey.g Then there are many incoming presence and messages, a couple of <r xmlns="urn:xmpp:sm:2" />. But it comes to connection reset.
  4. lovetox hm k, i have to test this myself and read me into smacks again, it doesnt tell me much now
  5. lovetox we are certainly not in the 1000
  6. lovetox do you know which version of ejabberd your server runs?
  7. andrey.g Unfortunately, no. Interesting, smacks acks come in batches of 3.
  8. andrey.g Some sort of bloated buffers?..
  9. lovetox ok it seems on older version the server default was 500
  10. Link Mauve Wasn’t there some plugin using the internal API that didn’t pass through the stanza counting mechanism?
  11. lovetox so 500 message on 15 rooms joined is possible
  12. Link Mauve I know using the XML console to send a stanza will mess up the count, for example.
  13. lovetox yeah but the console will tell you that
  14. lovetox it does not in this case, counts seem ok
  15. andrey.g The only enabled plugin I have is omemo.
  16. Link Mauve lovetox, it does?
  17. lovetox there are errors printed if the counts of server and client differ
  18. lovetox thats why you know that the console messes the count up
  19. Link Mauve lovetox, but errors are only printed if you open Gajim from the terminal, right?
  20. lovetox thats what andrey.g does .. i dont know what you are trying to say
  21. andrey.g After joining finished and all works, requests/acks come close together:
  22. andrey.g 10/03/2017 00:24:47 (I) nbxmpp.client_nb raising event from transport: :::::DATA RECEIVED:::: _____________ <r xmlns='urn:xmpp:sm:2'/> _____________ 10/03/2017 00:24:47 (I) nbxmpp.transports_nb Plugging fd 25, W:False, R:True 10/03/2017 00:24:47 (I) nbxmpp.client_nb raising event from transport: :::::DATA SENT:::: _____________ <a xmlns="urn:xmpp:sm:2" h="1290" />
  23. Link Mauve lovetox, that it should be explicit that using the XML console will mess up your sm state, or that using the XML console shouldn’t mess it up (such as by sending stanzas through the common APIs).
  24. lovetox Link Mauve, ok yeah, but totally different topic now, how is this helping andrey.g
  25. lovetox andrey.g, lets leave it for today, i have to reproduce this myself and try to find out whats wrong
  26. Link Mauve lovetox, maybe something else inside of Gajim is doing the same thing.
  27. Link Mauve That’s the best guess I can make currently.
  28. lovetox but thats easy to find out, as we can request an ack from the server, and get the count, and can compare it with ours
  29. lovetox and we do this, and smacks module sends a error message if doesnt fit
  30. lovetox so it was also my inital guess, but it seems not very likely now that i thought about it
  31. lovetox its more likely that for whatever reason gajim is to slow on sending acks
  32. lovetox and the server has a low value for unacked stanzas
  33. andrey.g OK. I see after 20 requests from Gajim, only 3 acks from server in batch, then again Gajim receives data and sends requests.
  34. andrey.g The problem with many "<r xmlns='urn:xmpp:sm:2'/>" is that there are no corresponding counters to follow.
  35. lovetox i think this stanzas are not counted
  36. Link Mauve This isn’t a stanza.
  37. lovetox nonzas
  38. lovetox :)
  39. Holger I think you guys were mixing up directions above. ejabberd's 0198 queue size limit is about stanzas sent from the server to the client. I.e. Gajim's outgoing count and the 'h' returned by the server are irrelevant.
  40. bot Philipp Hörist pushed 2 commits to branch _refs/heads/master_ of _gajim_ <>: *ef748c6d* <> Remove send button *0b83a675* <> Make Groupchat Roster larger
  41. bot Philipp Hörist pushed 2 commits to branch _refs/heads/gtk3_ of _gajim-plugins_ <>: *588e101f* <> [httpupload] Adapt to UI changes in Gajim *18a5477f* <> [httpupload] Update manifest.ini
  42. Holger And I doubt it's a counting issue. I mean this would only happen if the server's outgoing count and Gajim's incoming count differ by more than 500/1,000 stanzas. Seems unlikely.
  43. Holger ... unless maybe there's a Gajim bug that leads to Gajim no longer incrementing the incoming counter at all, or something.
  44. Holger Seems more likely to me that there simply *is* a stanza flood that Gajim fails to acknowledge in time.
  45. andrey.g I can be wrong, but what I see Gajim acknowledges received smack reqs fast, but floods(?) the servers itself with requests, which the servers acks only delayed. This is what I see in the console, since it's a SSL connection, it would take an effort to get setup to see real timings.
  46. andrey.g But enough for today, let's look at this later.
  47. bot Philipp Hörist pushed 5 commits to branch _refs/heads/gtk3_ of _gajim-plugins_ <>: *bb844593* <> Remove unused var *d79a211b* <> [tictactoe] Adapt to UI changes in Gajim *800aed51* <> [tictactoe] Update manifest.ini *a56f5381* <> [whiteboard] Adapt to UI changes in Gajim *b0af84b2* <> [whiteboard] Update manifest.ini
  48. Holger andrey.g: Do you have access to the server (logs)?
  49. Holger andrey.g: What Gajim version are you using?
  50. andrey.g Holger, unfortunately no: it's the super secret In the night of 29-30 September they did some maintenance work: after I received the warning message, there were only a couple of seconds to finish (I'd expect a little bit more). I'm on gajim_0.16 latest commit.
  51. bot Philipp Hörist pushed 1 commit to branch _refs/heads/master_ of _gajim_ <>: *7bd73e24* <> Fuse Emoticons/Authentication with TextView
  52. lovetox there you go cippaciong , tell me what you think :)
  53. lovetox damn missed the word Buttons in there ^^
  54. bot Yann Leboulanger pushed 1 commit to branch _refs/heads/master_ of _gajim_ <>: *6d10a5e2* <> fix translation strings
  55. lsg It can be fixed?
  56. debacle UI issue in master from 2017-10-01: When I change the style theme from "default" to another one, e.g. "human", "green", etc. the header colour in all conversations changes. But when I switch back to "default", the colour of last selected non-default theme remains.
  57. debacle Minor error: 16x16/status/channel-secure-croped-symbolic.symbolic.png should be 16 by 16 pixels, but is 16 by 11.
  58. lovetox debacle, why is that an issue?
  59. lovetox we need it 16 by 11
  60. lovetox i could have it 16 by 16 in the repos, but then i have to do a crop everytime i show the icon which seems wasteful
  61. lovetox or is the problem that its in the 16x16 folder
  62. debacle lovetox: Debian tool lintian warns about this. If it is not an accident but on purpose, no problem. I can just shut-up this warning :~)
  63. debacle lovetox: I just thought it was accidental.
  64. lovetox im on youtube and they play a 5 sec ad, that contains only the youtube logo and a jingle