Gesendet: Donnerstag, 02. Oktober 2014 um 18:17 Uhr Von: "Stuart Henderson" <s...@spacehopper.org> An: misc@openbsd.org Betreff: Re: cvs checkout: Corrupt MAC on input On 2014-10-02, Stefan Wollny <stefan.wol...@web.de> wrote: >> Hi there! >> >> This morning I have had to reinstall my squid-server running >> amd64-current from scratch (made a dump error...). >> >> OpenBSD 5.6-current (GENERIC.MP) #394: Wed Oct 1 12:54:54 MDT 2014 >> dera...@amd64.openbsd.org:/usr/src/sys/arch/amd64/compile/GENERIC.MP >> (Full dmesg at the end) >> >> When getting the sources from ftp.hostserver.de I noticed s.th. unpleasant: >> >><quote> >> U src/gnu/gcc/gcc/tree-ssa-forwprop.c >> Corrupted MAC on input. >> Disconnecting: Packet corrupt >> cvs [checkout aborted]: end of file from server (consult above message >> if any) >></quote> > > This implies a packet that was corrupted, but still had a valid TCP > checksum (otherwise SSH wouldn't have seen the packet at all). > > This can happen if a NIC has TCP checksum offloading, so it verifies > the packet was OK at receipt, but there is some corruption between > being received by the NIC and being processed by software. > > The most probable cause in that case is a hardware problem, though to > see it on two different hosts at the same time would be highly unlikely. > > Another possibility might be if some router/nat box is somehow breaking > the contents of packets and regenerating TCP checksums. > > Are you aware of any changes to hardware that might be common in the > network path between both your squid server and laptop?
Hi Stuart, sorry for replying late - I was off for a long weekend. To answer your question - no, there have been no changes at all. It is the network in my home office so I know for shure. Last night I reinstalled the laptop from the latest snapshots (#394) and to rule out some kind of 'hickup' of the cable-modem, the router or the switches I power-cycled every instance. Even though I didn't see the error message originally reported I still was not able to get the source-tree by cvs: After a few seconds the system looses its network, entirely. A non-technical description of the impression I got is "the system can't take the load by the sheer number of files received". (Of course I know this is not what a bug report should sound like... ;-) ) Calling 'netstat' or 'route' shows ... nothing! Both report the lack of any routes. This only happens with cvs, not with 'pkg_add'! The system won't reattach to the network by '/etc/netstart', I have to reboot to get the routes back up. As others have reported issues with dhclient recently a first 'shot in the dark' would be that this is where a suspect is living... I am not shure what to report exactly tonight as I am off from home in a boring hotel room. E.g. if the routes are lost again: What should I report? Any hints? Thank you! STEFAN
