View previous topic :: View next topic |
Author |
Message |
Ktyak n00b
Joined: 25 Jun 2005 Posts: 10
|
Posted: Sat Jun 25, 2005 11:28 pm Post subject: [SOLVED] Random packet loss, rsync failures |
|
|
Hi,
I've spent the last couple of days trying to install Gentoo 2005.0 (with LVM2) on my new Epia SP 13000.
I'm working my way through the Handbook and have (mostly) a working network connection, partitioned the drives
and am at the point where I am trying to "emerge -sync".
Now, I've posted this in the network section because that seems to be where my problem lies.
I'm not too familiar with rsync but when emerge is trying to get a list of files (theres a counter on the screen that increases) - it reaches anywhere from say 4000 files to 104000 (the highest count I managed to reach) - and then the counter stops and evertually the rsync mechanism times outs.
I tried incresing the timeout but that made no difference, apart from making me wait longer before the timeout ocurred.
I tried different rsync servers and get the same problem.
Now, prior to to the emerge operation I had to use links2 to download the stage1 and portege tars and it took me ages because the downloads would freeze - eventually I succeded by manually doing a "wget -c" repeatedly until I was able to completely get each file.
I have a local lan, with a static block of 8 ips, the other 'windows' machines on my lan are able to get out on to the internet, download large files, trace route and ping to external sites (including the ones that the linux box is trying to get to) without any packet loss or problem.
But, the gentoo install, altho able to access machines on my local lan and remote sites, reports random packet loss if i try to ping to any site - this is why I said the network is *mostly* working
I'm at a bit of a loss what to try next and would appreaciate any suggestions.
Thanks
KT.
Last edited by Ktyak on Mon Jun 27, 2005 9:00 am; edited 1 time in total |
|
Back to top |
|
|
think4urs11 Bodhisattva
Joined: 25 Jun 2003 Posts: 6659 Location: above the cloud
|
Posted: Sun Jun 26, 2005 6:32 am Post subject: |
|
|
Hi,
beeing owner of a CL6000 board i'd the same issues.
1. These boards (especially the VT6105 NIC) don't like ACPI so i'd suggest to turn it off.
2. Did you check your speed/duplex settings and tried to use fixed values instead of auto negotiation?
HTH
T. _________________ Nothing is secure / Security is always a trade-off with usability / Do not assume anything / Trust no-one, nothing / Paranoia is your friend / Think for yourself |
|
Back to top |
|
|
Ktyak n00b
Joined: 25 Jun 2005 Posts: 10
|
Posted: Sun Jun 26, 2005 9:57 am Post subject: |
|
|
Hi,
Thanks for the suggestions.
I think I know how to disable ACPI but as I'd rather not reboot the linux box just yet I'll try the nic speed/autosense settings first but I'm not sure where to find those.
Any pointers ?
KT. |
|
Back to top |
|
|
think4urs11 Bodhisattva
Joined: 25 Jun 2003 Posts: 6659 Location: above the cloud
|
Posted: Sun Jun 26, 2005 10:19 am Post subject: |
|
|
ethtool and/or mii-tool are your friends _________________ Nothing is secure / Security is always a trade-off with usability / Do not assume anything / Trust no-one, nothing / Paranoia is your friend / Think for yourself |
|
Back to top |
|
|
Ktyak n00b
Joined: 25 Jun 2005 Posts: 10
|
Posted: Sun Jun 26, 2005 11:56 am Post subject: |
|
|
Thanks again.
Ok, I managed to disable autonegotiation and forced the nic to use 10Mb Full duplex using eth-tool but unfortunately the packet loss issues continued.
So, I've just rebooted from the live cd and specified acpi=off as a boot parameter to the gentoo kernel and once in, using lsmod does not show anything for acpi and there was a message on startup to say that the acpi interpreter was disabled so I'm assuming acpi is definately off.
I've also checked again that auto-neg is still off and that the speed is fixed at 10Mb.
Still no luck. If I ping an external site from the linux box I average between 60-70% packet loss but get no packet loss from other boxes on the local network.
....
But!! just as I was about to finish this post I decided to check the local nets router config again and it struck me as odd that it was configured to use the network address as its address so I changed - eg the network address ranges were x.x.x.96 to x.x.x.103 - the router was listening on .96 internally as well as .102
Some how it was working for everything else but not for the linux box.
I changed it so that it was only listening on .102 and suddenly the random packet loss disappeared.
How it worked at all from the windows boxes without the loss I can't say.
By virtue of DHCP the windows boxes were using .96 as a gateway address but it should have been .102 as I had entered into the linux box.
Weird.
Anyway many thanks for the help, I was pulling what little remains of my hair out.
KT. |
|
Back to top |
|
|
|