-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA256 retitle 612426 multistrap: package configuration sequence variations quit
On 08/02/11 17:19, Johannes Schauer wrote: > hi, > > i have my multistrap config and the output of a run attached. As you can see > it is version 2.1.7 but using 2.1.10 results into the same problem. I don't get that problem. Log attached. > as you can also see in the multistrap log in constrast to your log base-files > is > configured before base-passwd which makes the configuration of base-files > fail like this: > > Setting up base-files (6.1) ... > chown: invalid user: `root:root' > dpkg: error processing base-files (--configure): > subprocess installed post-installation script returned error exit status 1 That needs further investigation because it doesn't happen at this end. Currently, unpacking is tied to configuration in native mode - if you know your perl you can decouple it around line 359 of multistrap. That would allow you to chroot into the unpacked directory and try dpkg - --configure -a yourself. Also make sure you remove the entire tree between tests - multistrap will pick up where the last one left off but if that wasn't the same configuration, things will go wrong. > and in the end i get: > > Errors were encountered while processing: > base-files > bash bash is more interesting there. That is probably an incidence of a bug fixed in experimental during the freeze. You could try with multistrap 2.1.7 which has just arrived in unstable. The preinst is being called with the upgrade argument when new versions use install. You shouldn't need to use noauth=true either - set that to false before trying 2.1.7 (another bug fixed during the freeze). > the whole process still exits with return code 0 but this is not the case > when i > do a multistrap for a foreign architecture and then do `dpkg --configure -a` > manually. What do you mean by foreign (if it is foreign, you can't execute programs within the chroot)? > dpkg will then fail with a non-zero exit status with the same output as above. > It is just that when running multistrap with a native architecture, then it > doesnt > check the return code of `dpkg --configure -a` and even though it fails (at > least > here) the whole process doesnt. I'm going to change that and let multistrap accumulate non-fatal errors (like this one) and report them appropriately (including multistrap itself exiting with the number of warnings detected). This will be extended to the hook support too (new to 2.1.7). - -- Neil Williams ============= http://www.linux.codehelp.co.uk/ -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.10 (GNU/Linux) Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/ iQIcBAEBCAAGBQJNUaGFAAoJEPFn5DyBQ7aCpKwP+gJOV/x2rdHGyHU5kxhsWvp/ kCPR+sRo+ySwWzxIZvnUPWcPux106E0Wbwzlc7nXntWurrnP2rqnwSLB7/3tTvoC /IiIpRQERFUYN7F1EfN3UcjAnY7Uoq2cCbXDWxuA9XVhCHHpBnVCLGySxVQDLgXb 2AVHeMDrf+gJ5SO8TS+9SiqcPHONbQaahz9wSsRV4alZkolpxY050en9nOtWicwS 4VO1ZRrMjZ0VjB9SR1lhLuHvBj4kBXkI2TsXjfqfUytULetN2RZEPr+Aw90+Dm7H ICJhnvPt4kY1n5a4HUYAnOryxpDiLCGg5k6JstkUz2dJzhmmunZt7QtCQtO0CbLa H4mg8h1K/yWBd6EP9wrvgUM+XI4qCZsTw2qOnTCUdBVNJOVE5bfuXrZPWQsQzT8x 8afWpAHq+iMKx6C5h6YkrnTc5XPfaNGkiRCl7JPr0DSU6+C6fxHwX62DwI9fgMrm 6ZaNIVssA3oyj+olnW2MA13Z3XK84tvj+kR45n5/JEAjHCrxjKUMt28p3KWLGUyA SpBXQpHEtYkcJ9WSvmanZGECIA1nAWyymgePyKecUkBKt0gfDVi0oulHYSo9LKLe WZXFq0IqVAC+TTvbtHYBdaCWV7tk09pBiX90xfsz3FSl3E8LHieRXaWNSgfKUyLh Xwzxai49+b5c7hIT2BHt =Nw3c -----END PGP SIGNATURE-----
unstable.log.gz
Description: GNU Zip compressed data
