SWAP causes connection loss
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
ltsp (Ubuntu) |
Fix Released
|
Undecided
|
Oliver Grawert |
Bug Description
My first bug report.
Gutsy 7.10 Host. Gutsy 7.10 LTSP Client Chroot.
When booting an LTSP client, using NBD and squashfs (Default in LTSP 5+), I kept getting an error where the squashfs volume would just disappear.
Traced it down to the script:
/etc/
When enabling SWAP over nbd, all hell breaks loose. For example, I have the following in the clients /boot/lts.conf (which is copied to the correct place when the host executes ltsp-update-
---
[default]
SOUND=True
LOCALDEV=False
NBD_SWAP=True
NBD_
NBD_PORT=9572
SYSLOG=server
---
The reason is that the SWAP nbd is using /dev/nbd0. So is the currently mounted /rofs which is a component of the / unionfs.
The fix is to change the file to use /dev/nbd1.
- Brendan
you can drop the NBD_PORT and NBD_SERVER directives both are set automatically ... i386/etc/ init.d/ ltsp-client- setup is hardcoded to nbd0 ...
actually the device in /opt/ltsp/
you can work around it by changing teh device name on line 65 and 66 (and run ltsp-update-image afterwards)
i'll work on a proper autodetection mechanism that automatically takes the last free device for hardy (8.04)