Hi!
As the title sais, I’m having troubles with Spark 2.6.3 and openfire 3.7. Everything was working fine till the day I restarted the machine where Openfire was installed. From that day, any transfer stops at 256MB and the speed never goes over 1mb/sec (it should be 10mb/sec) If I reinstall Spark and restart the computer everything works fine again (speed raises to 10mb/sec and I can transfer huge files, even more than 14GB) but four or five days later, the speed goes down again and I can’t transfer files larger than 256MB.
2011.09.09 12:36:09 Error or result packet could not be delivered http://jabber.org/protocol/bytestreams
That’s what the log sais when it crashes.
I have several backups of openfire, so I went back to one wich I know it was working right. But nothing changes.
I’ve everything up to date, so right now I’m lost. ¿Why 256MB exactly? ¿Why reinstalling it works for a couple of days? ¿Any hint?
Looks like your OF server is acting as file transfer proxy. That’s a setting in the server console. If it us not set up as file transfer proxy I would be suprised.
File transfer in 2.6.3 has a couple of issue. These will be fixed in 2.7.0. Due in the next weeks.
It’s not set to act as a proxy. ¿Should it be?
Hi
no, it should not be set as file transfer proxy. The issue you are reporting look like a client that uses IBB as file transfer mechanism. This can be seen in the debug log by send/receive packets that contain large text chunks. The IBB transfer is broken in 2.6.3 and will be fixed in 2.7.0 for all clients of the 2.7.0 range and above.
Walter
From “Raw sent packets”:
<iq id="69y86-131129" to="user2@galiframe/Spark 2.6.3" from="user1@galiframe/Spark 2.6.3" type="error">
<data xmlns="[http://jabber.org/protocol/ibb](http://jabber.org/protocol/ibb)" seq="0" sid="jsi_3909734181244486510">ALotOfDataHere</data>
<error code="400" type="WAIT">
<unexpected-request xmlns="urn:ietf:params:xml:ns:xmpp-stanzas"/>
</error>
</iq>
From “Raw received packets”:
<iq id="69y86-131130" to="user1@galiframe/Spark 2.6.3" type="set" from="user2@galiframe/Spark 2.6.3"><close xmlns="[http://jabber.org/protocol/ibb](http://jabber.org/protocol/ibb)" sid="jsi_3909734181244486510"/></iq>
I saw those large packets in the debug mode of spark so it’s IBB for sure in all clients. So ¿Should I revert all spark clients to previous versions till the 2.7.0 version of Spark is released? Where can I found older versions? Wich one is recommended since we’re going to transfer huge files via Spark?
Thanks for the info!
An unexpected-request error occurs if the sequencenumber of the packet was already received in a previous packet. We have noticed this while hunting for the IBB filetransfer bug.
This bug is a bit tricky so we still working on it. We’ ve uploaded a alpha-smack which will use message stanzas instead of iq stanzas to bypass this problem, while we try to fix the bug iq stanza mode.
You could try that smack, maybe it will solve your problem until the new spark is released.
Anyway, the IBB mode, as a backup for Bytestream, is not that fast. So the transfer-speed is lower then 1mb/sec. The question is, what causes your Bytestream to fail.
Can you post the send and received packets while filetransfer for me (until first occur of a packet with “ALotOfDataHere”)?