summaryrefslogtreecommitdiffstats
path: root/perl-install/raid.pm
Commit message (Collapse)AuthorAgeFilesLines
* use --force when creating a md with only one device (bugzilla #21214)Pascal Rigaux2006-02-201-1/+3
|
* we shouldn't rely on callers to load missing modules...Thierry Vignaud2005-12-221-0/+1
|
* "mdadm --detail --brief" doesn't contain "devices=..." anymore (since mdadm ↵Pascal Rigaux2005-10-181-1/+1
| | | | | | | 1.12.0), it needs option "-v" to keep previous behaviour
* write /etc/mdadm.conf when creating a new md (bugzilla #15502)Pascal Rigaux2005-08-081-1/+2
|
* - move functions using /proc/partitions out of fsedit to fs::proc_partitionsPascal Rigaux2005-06-101-1/+0
| | | | | | - remove unneeded "use xxx" - add some "use xxx" (nb: not completly needed because some other modules may do it)
* don't need prefixing with current packagePascal Rigaux2005-06-091-1/+1
|
* remove unused codePascal Rigaux2005-06-031-4/+0
|
* calling inactivate_and_dirty() on a new structure is bad, we loose the ↵Pascal Rigaux2005-03-241-4/+5
| | | | | | | {isFormatted} flag (no big deal though, it happened because raid::new() was creating a new raid with an already active md name)
* ensure we use/propose a free md when creating a new onePascal Rigaux2005-03-241-1/+6
|
* after "mdadm --assemble" there can be some mds in inactivate state busying ↵Pascal Rigaux2005-03-241-1/+9
| | | | devices, stopping them
* switch back to auto=yes :Pascal Rigaux2005-01-271-1/+1
| | | | | | | | | | | | | | | | | | | | > > the only problem with auto=yes is that it ignores the minor number > > specified and always uses the first avaliable minor number, so you might > > find /dev/md2 with minor 0, /proc/mdstat will call it based on minor > > number. This is going to get really confusing for users. > > I was planning to fix this in upstream source, but i have not been able > > to find time to do it yet. > > ok, i fixed that in mdadm-1.8.0-2mdk > there is a new option (auto=dev) that will create the device file with > the correct minor number based on the device name. > so please put auto=dev instead of auto=yes in the /etc/mdadm.conf > please hold the auto=dev change, the mdadm author feel this should be the default behaviour for auto=yes, so i will upload a new mdadm-1.9.0-0pre1.1mdk with fully working auto as soon as i return to italy (this weekend i believe)
* fix "mdadm package not installed" during install (bugzilla #13259)Pascal Rigaux2005-01-271-1/+1
|
* put the new option auto=dev instead of auto=yes (see mdadm-1.8.0-2mdk)Pascal Rigaux2005-01-231-1/+1
|
* - don't write /etc/mdadm.conf when no raidPascal Rigaux2005-01-211-1/+3
| | | | | | - use option "auto=yes" in mdadm.conf to ensure mdadm will create /dev/mdX devices when needed (those are not there when using udev and neither kernel raid autostart nor initrd created the raid)
* - fs::format::part_raw() now takes $wait_message to allow displaying the ↵Pascal Rigaux2005-01-061-1/+1
| | | | | | | | progress of format - create fs::format::wait_message() which creates a $wait_message valid to give to fs::format::part - fs::format::mke2fs() format while parsing the output to display the progress
* better english (writing style rather than spoken one)Thierry Vignaud2004-12-131-1/+1
|
* remove some unneeded ";", add some for normalization (as told by perl_checker)Pascal Rigaux2004-11-181-1/+1
|
* since we need mdadm, ensure we have it (bugzilla #12146)Pascal Rigaux2004-10-271-0/+5
|
* newly created raids must have a fs_type (this was dropped in rev 1.45, it ↵Pascal Rigaux2004-10-261-0/+1
| | | | was an error)
* put UUID instead of devices in mdadm.conf to be more device naming independantPascal Rigaux2004-10-141-13/+22
| | | | | (as requested by Luca Berra on cooker)
* saferPascal Rigaux2004-09-061-2/+3
|
* fix typoPascal Rigaux2004-09-061-3/+4
|
* handle pt_type, not only fs_typePascal Rigaux2004-09-061-2/+5
|
* be more failsafe with half broken existing raidsPascal Rigaux2004-09-061-0/+1
|
* remove dead codePascal Rigaux2004-08-101-4/+0
|
* - switch to mdadm (instead of raidtools)Pascal Rigaux2004-08-081-76/+128
| | | | | | | | | - create mdadm.conf instead of raidtab - internal {raids} is no more indexed by X for mdX, and so don't have holes anymore - internal {chunk-size} is now a number in KiB - internal {raid} is the raid device name, not the number - various cleanup for raid detection
* - add field {fs_type} partially replacing {pt_type}Pascal Rigaux2004-07-271-5/+5
| | | | | | | | | | | | | | | | | | | | {pt_type} is always a number, {fs_type} is always a string - introduce set_isFormatted() (to ensure {notFormatted} but also {fs_type_from_magic} and {bad_fs_type_magic} are updated) - don't use 0x483 for ext3 anymore (same for reiserfs...), the type_name gives both a pt_type and a fs_type - many accessors from partition_table removed (type2fs(), fs2pt_type()) - remove isThisFs() (not useful anymore since we can use directly {fs_type}) - remove isFat() (inline the function) - other isXXX() from partition_table are moved to fs::type - part2name() is now fs::type::part2type_name - name2pt_type() is now fs::type::type_name2subpart() - partition_table::important_types() is now fs::type::type_names() - fsedit::typeOfPart() is now fs::type::fs_type_from_magic() - no need to truncate type_name since they are shorter
* - move format related functions out of fs.pm to new module fs/format.pmPascal Rigaux2004-07-231-1/+1
| | | | | - remove swap.pm, moving its few functions to fs/format.pm or fs.pm
* big renaming of field {type} to {pt_type},Pascal Rigaux2004-07-051-1/+1
| | | | | | this will allow defining {fs_type} which will always be a string whereas {pt_type} will always be a number
* perl now handle cleanly utf8 string in exceptions, no need to die \N("...")Pascal Rigaux2004-04-051-2/+2
|
* perl_checker adaptations + fixesPascal Rigaux2003-04-241-2/+2
|
* to workaround perl bug removing UTF8 flag when passing scalars to die's, passGuillaume Cottenceau2003-02-191-2/+2
| | | | | | a scalar-ref. but we need to de-ref, so it might break many things :). let's make a prayer :).
* mkraid wants all the md devices written in raidtab to exist, even if asking toPascal Rigaux2003-02-191-2/+2
| | | | | create a specific md
* cleanupPascal Rigaux2003-02-191-2/+6
|
* perl_checker compliance ("ref" now need parentheses in many case)Pascal Rigaux2003-02-121-1/+1
|
* software raid (mdX) can go up to md31Pascal Rigaux2003-01-291-1/+3
|
* use "if any" instead of "if grep", and various other occurences of "any", ↵Pascal Rigaux2003-01-071-1/+1
| | | | "every", "partition"
* use output()Pascal Rigaux2002-12-031-14/+13
|
* testing awareGuillaume Cottenceau2002-11-281-0/+1
|
* various small syntax enhancements to please perl_checkerPascal Rigaux2002-11-111-1/+1
|
* - have "local *FILEHANDLE" before each "open FILEHANDLE, ..."Pascal Rigaux2002-11-061-1/+1
| | | | | - use some "cat_" and "output" where possible
* replace "_" with "N" and "__" with "N_"Pascal Rigaux2002-11-061-4/+4
| | | | | | | | | | | | | | | | | | | rationale: - currently, we use _("xxx") as a shorthand for gettext("xxx"). It also used to call xgettext with --keyword=_ - alas, function &_ is global and not by package (notice esp. that _ is not exported in common.pm) - this lead to big ugly pb with packages defining their own &_, overriding common.pm's &_ - a fix is to set @::textdomains to add a new domain (the default being "libDrakX") but relying on the global "_" is still dangerous!
* use each_index instead of map_index when the return value is not usedPascal Rigaux2002-11-051-1/+1
|
* also create a raidtab if it doesn't exist (occurs when / has been formatted)Pascal Rigaux2002-09-241-0/+2
|
* make new perl_checker happy (and that's not easy!)Pascal Rigaux2002-07-311-4/+4
|
* ensure the raid is disactivated and marked dirty on any actions that destroy ↵Pascal Rigaux2002-07-071-14/+26
| | | | the raid
* defaults to ext3 instead of ext2Pascal Rigaux2002-01-221-1/+1
|
* use new mkdir_p, rm_rf and cp_af from MDK::CommonPascal Rigaux2001-09-161-2/+1
|
* update the raid size as it should bePascal Rigaux2001-08-181-1/+3
|
* fixed i18nStefan Siegel2001-08-181-1/+1
|