21:07:53 good evening. small problem here. i have an ancient server running (# uname -a   SunOS kvm1 5.11 omnios-b281e50 i86pc i386 i86pc) where the zpool was having problems. tried to change a disk, didnt work because of wrong sector alignment or whatsoever. made a new pool, send the data and then ... tried to export the pool and import it under the old 21:07:53 name. and there i ran into this bug: https://github.com/illumos/illumos-gate/commit/bd0f709169e67f4bd34526e186a7c34f595f0d9b . so i can't reimport the pool anymore, and i fear the system is too old to do any updates. any chances to replace the zpool binary or do something else nasty? 22:11:08 # zpool import tank datapool 22:11:09 ld.so.1: zpool: fatal: libc.so.1: version 'ILLUMOS_0.37' not found (required by file /sbin/zpool) 22:11:09 ld.so.1: zpool: fatal: libc.so.1: open failed: No such file or directory 22:11:10 replacing with a current version doesnt work (as i would have expected) 22:11:32 so maybe i can try to grab an old installation image? 22:12:41 Your `zpool` binary needs a more modern libc than what's in your library search path. 22:13:17 Did you just grab a replacement zpool binary? 22:13:25 yeah, just because i was curious 22:13:27 If so, you'll need to pull along a matching libc for it. 22:13:46 i fear i'll brick the rest of the system when i replace the libc 22:13:54 you oculd do this. 22:14:09 - put a matching libc in /tmp/ 22:14:36 (it'll need to be from /lib/amd64/libc.so if zpool is 64-bit. `file /sbin/zpool` is your friend) 22:14:54 - run zpool using LD_LIBRARY_PATH overrrides. Assume you have the matching libc.so in /tmp: 22:15:04 (and you're using bash or sh or ksh) 22:15:11 ksh on that one 22:15:24 LD_LIBRARY_PATH=/tmp /path/to/new/zpool import 22:15:32 you MAY need other libraries too? 22:15:57 This may cause more problems if it's a dependency hell thing. 22:16:10 i think so. maybe i'll look for an old system image of a slightly newer version. 22:16:24 if anyone has something around :) 22:17:06 Sorry. The other trick I can think of involved running 'zpool' in mdb, and editing out the assert() call. :) 22:17:37 * danmcd has binary patched before, so he's biased. :) 22:17:46 sounds good to me. but i guess thats way above my level :)