Note: "permalinks" may not be as permanent as we would like,
direct links of old sources may well be a few messages off.
RobH wrote: > > Hi, > > Yesterday I have tried to get DRBD 0.7.5 work on a RHEL 3 machine the > system crashed while syncing, 30 seconds after starting it. > > What I have done now is to Compile DRBD 0.6.13 but this doesn't work > also. > Kernel is 2.4.21-15EL Compile when with out any problems, have use the > described Red hat procedure to compile drbd . > > 1. go to /boot and cp config-2.4.xxxx /usr/src/linux-2.4/ works better to cp /boot/config-2.4.xxxx /usr/src/linux-2.4/.config as .config is what gets used. > 2. change to /usr/src/linux-2.4/ > 3. make mrproper You just blew away your config file, not nessary, i.e., DON'T DO THAT :). > 4. vi Makefile and remove the custom text Actually you would have been better leaving the custom text or changing it to something like RobH1, as now you have caused the process to write over the modules redhat had installed for you (not impossible to go back, just not easy). > 5. make -s oldconfig_nonint > 6. make dep with redhat I would suggest (note: my experience comes from fedora core 1&2 so it might be different in RHEL) building an rpm and installing it. replace your 7 & 8 with 7. #assuming RHEL 3, like fedora core 1, uses gcc32 to make its kernels instead of gcc export CC=gcc32 #you should find out about this from other RHEL people 7.5 make rpm 8. #install new kernel and modules rpm -ivh \ /usr/src/redhat/RPMS/i386/kernel-2.4???RobH1-1.i386.rpm 8.5. #update build initrd and update grub /sbin/new-kernel-pkg --mkinitrd --depmod --install \ 2.4.???RobH1 <SNIP> Now you need to reboot, and select your new kernel from grub. You do this because drbd's build process looks to the running kernel to figure out what to build against. > 9. tar -xvzf drbd-0.6.13.tar.gz to /usr/src > 10. change to /usr/src/drbd-0.6.13/ > 11. vi drbd_config.h and edit the #define SIGHAND_HACK > 12. change to /usr/src/drbd-0.6.13 replace steps 13-15 with: 13. make rpm 14. rpm -ivh dist/RPMS/i386/drbd-* > <SNIP Log> do us another slight favor, sync the clocks between the two nodes. The logs look like one was seeing the occurrences ~1 hour later than the other. It is hard to reconcile the occurrences with that big a delta. I like NTP for syncing mine. If you don't have a local ntp source you could set one of the boxes ntp to fudge a local clock, and have the other sync to it. assuming cluster01 is 10.0.0.1 cluster02 is 10.0.0.2 in ntp.conf #on cluster02 add the lines server 127.127.1.0 # local clock fudge 127.127.1.0 stratum 10 restrict 10.0.0.1 mask 255.255.255.255 notrap #on cluster01 add the lines restrict 10.0.0.2 mask 255.255.255.255 notrap server cluster02 maxpoll 11 burst -- Todd Denniston Crane Division, Naval Surface Warfare Center (NSWC Crane) Harnessing the Power of Technology for the Warfighter