Opened 3 years ago

Closed 2 years ago

#17777 closed bug (fixed)

Networking stops working after a while from boot

Reported by: Nexus-6 Owned by: axeld
Priority: normal Milestone: R1/beta4
Component: Network & Internet/Stack Version: R1/Development
Keywords: Cc:
Blocked By: Blocking:
Platform: All

Description

I've been experiencing some troubles with networking for 2/3 weeks now. After a while networking stops working, no inbound/outbound traffic is performed by the ethernet adapter (82579LM Gigabit Network Connection (Lewisville)). It could be minutes or even hours but in the end I must reboot. The ethernet card appears to be fully functional in the network preference panel, though. Last incident occurred today around 13:50 CEST (please see syslog attached). AFACT the latest working build was hrev56129. There were two events happening in the same time frame, I started actively developing a new version of Weather to support OpenMeteo forecasting provider (there was a PR never merged to support MetaWeather, but as you may know it is now dead). The old code is not perfectly optimized, to say the least. I'm thinking of it breaking the network stack some way with abnormal or malformed packets? The second was the unfortunate event related to updating to hrev56131 that led me to restore my installation (see ticket #17759).

Not sure if there is any correlation but it's worth mentioning.

Attachments (1)

syslog (344.4 KB ) - added by Nexus-6 3 years ago.

Download all attachments as: .zip

Change History (4)

by Nexus-6, 3 years ago

Attachment: syslog added

comment:1 by waddlesplash, 3 years ago

Please paste the output of ifconfig when you get into this state: it may show a lot of errors.

I don't see anything of relevance in the syslog unfortunately, in any of the boots, related to the ipro1000 driver.

comment:2 by Nexus-6, 3 years ago

Since the last update to hrev56168 I have not experienced the problem any longer.

comment:3 by waddlesplash, 2 years ago

Resolution: fixed
Status: newclosed

It's possible the jumbo MTUs overhaul caused this, as it happened around the right timeframe a few days after hrev56129, and then the too-large-packet handling fix in hrev56164~1 resolved the problem.

In that case, please keep an eye on your syslog for errors that contain: "error reading packet: too large" -- if you see one of these after some time of use, please at least write a comment here with the syslog excerpt surrounding that, and I will determine whether it warrants another ticket.

Note: See TracTickets for help on using tickets.