| em22_ | hello, Im trying to install ubuntustudio beta1. I used the "Something Else" option. everything was going fine. I made some choices, and then changed my mind and reverted them. then I quit the installer. looked at the disk. then started over. now the installer won't go past the initial screen -- it is the point where it detects if any partitions are mounted or in the general area where that happens. I freaked thinking the | 18:37 |
|---|---|---|
| em22_ | disk was damaged, that the revert had not worked. but, no, everything boots fine, I have about a dozen partiotns and about 5 versions of linux. I booted my main system and also the system I intended to replace and had made the reverted changes to. I was doing an install from flash, so then I thought, maybe the flash saved a setting which was causing this problem, so I tried a DVD and got the same result. this is really strange | 18:37 |
| em22_ | . program is not frozen, when I choose quit it responds just fine. but after clicking continue at the screen where it asks if you want to also install mp3 etc. then the cursor just spins forever. any ideas? | 18:37 |
| xnox | em22_: bug 1080701 | 19:14 |
| ubot2` | Launchpad bug 1080701 in ubiquity (Ubuntu Raring) "After 'Preparing to install Ubuntu' screen, raring installation hangs" [High,Confirmed] https://launchpad.net/bugs/1080701 | 19:14 |
| em22_ | xnox: Thank You! yes, that is the bug alright. I will try the suggested work-arounds and see what happens | 19:55 |
| infinity | xnox: Do you have a fix in the works for that one before release? | 20:06 |
| xnox | infinity: i'm failing to reproduce it reliable and strace it at the same time to figure out why partman-* shell scripts are hanging and not returning =/ | 20:16 |
| xnox | s/reliable/reliably/ | 20:16 |
| em22_ | xnox: Im suspecious that the part table may be getting altered in some subtle way. what I encountered is that the firt couple of times I ran the installer -- and aborted... everything worked exactly as expected. then I had a gee I wonder what that does moment.... on the partition manager page, next to the Change button, there is a [+] and [-] without any explaination of what they are for/do. I guessed that they added and r | 20:41 |
| em22_ | emoved, but decided to try it, being falsely confident in revert and other prompts to fend off disaster. So it allowed me to delete sad11 on a disk with 13 partitions. this seems like a very bad idea to me, to create a hole in the middle of the chain should not have been allowed. so after seeing what it intended to do, I did a revert. it was after that the trouble started, everything had been fine until then. | 20:41 |
| em22_ | xnox: where are these scripts? I'd like to take a look at them. | 20:44 |
| xnox | em22_: /var/lib/partman/automatically_partition/[15reuse|25replace] | 20:49 |
| em22_ | a related issue, again pointing to the partition chain being altered in some way.... when you make a Change and specify the mount point and format type, then after hitting OK, it brings up a resize prompt, which declares that it is going to resize the disk and it may take awhile etc. But the thing is, I did not request to change the size of the partition, and it would break things if it was altered, so it is a bug that this dia | 20:49 |
| em22_ | log comes up. But I always cancel it... so no harm done, but other people may not cancel it.... not sure what it will do when resizing the partition to be the same size it already is... or maybe off by one. but something defintely rotten here. | 20:49 |
| em22_ | a further observation... my partitions are not track aligned. this is due to the original windows layout. | 20:50 |
| em22_ | xnox: thanks, I will take a look | 20:51 |
| infinity | em22_: There's nothing wrong with deleting a partition in the middle of the disk, I don't see why you think that should be disallowed. | 20:58 |
| ogra_ | its not very effective though | 21:01 |
| em22_ | because, even though it is allowed for the chain to point to partitions that are not in linear address sequence, most programs that access the partitons do assume that they are in sequence, this leads to disk corruption major bad personal experience with that one (cough cough windows). in addition to which, on a disk with multiple oses, if you start arbitrarly changing the partition numbers you break any os that is not using uuid | 21:08 |
| em22_ | to find them. it is only safe to make changes at the end of the chain, anything else is very high risk. | 21:08 |
| infinity | em22_: Erm, almost none of that is true. | 21:12 |
| infinity | em22_: Partition length are defined in the partition table, nothing just scans and hopes, at least, nothing written in the last decade or two. | 21:12 |
| em22_ | infinity: you do know about fstab dont you??? | 21:15 |
| infinity | Sounds vaguely familiar. Tell me more. | 21:16 |
| em22_ | you know... the place where partitons are typically specified by their sequence number... and if that number changes, then the entries are pointing to the wrong thing. try man fstab | 21:16 |
| infinity | em22_: That was sarcasm in response to your talking down to me, but thanks. | 21:18 |
| em22_ | well, I should hope that it would be sarcasm... but you seem to have missed the point that the partitions are often specified by number and adding or deleting partitons changes that number. and thus breaks any os expecting that number to be there. and yes, in a matter of speaking, oses do indeed scan the partition table... because anything more than 4 partitions is stared as a LINKED LIST which must be walked. windows expects | 21:24 |
| em22_ | this linked list to be in linear address sequence, really bad things happen if it is not. even though technically the list could be out of order... you will destroy your disk if you have any windows partitions on there... Ive had this happen so not theoory here. | 21:24 |
| ogra_ | the only thing removing a partition in the middle causes is latency for the heads since their ways get longer when jumping between inner and outer partitions (if you actually use two on the ends) | 21:43 |
| em22_ | ogra_: you assume far too much... try a set up with 5 different oses installed, to each os, add all of the not-auto-detected partitions manually, using their number, e.g. sdaXX. now delete one of the partitions in the middle and see what happens to your oses. I shall not make any further comments on this subject. further information can be found in the many fine documents. | 22:04 |
| * ogra_ wonders why anyone would have 5 OSes installed .... | 22:06 | |
| em22_ | people who like to try different things and people who develop software, both categories are likely to install multiple oses. | 22:08 |
| cjwatson | none of this means that the facility should be absent, since many fine and sensible operating systems (e.g. Ubuntu) use UUIDs in fstab and really don't care about being renumbered. at most perhaps it justifies a warning. | 22:08 |
| cjwatson | anyway I'm sceptical that this is related; I'm pretty sure the transactional properties of partition manager operations are working | 22:10 |
| em22_ | a warning would be good, an ADVANCED button would be better. only the boot partiton and the swap partition are auto configured for UUID. when adding the other partitions manually it is more convenient to use the number. | 22:10 |
| cjwatson | any partitions that you configure while installing Ubuntu will use UUIDs | 22:11 |
| cjwatson | we won't be adding a separate advanced button | 22:11 |
| cjwatson | they're pernicious warts on UI | 22:11 |
| cjwatson | in any case: the usual things minimally necessary to debug this kind of problem are /var/log/syslog and /var/log/partman from the running installer after reproducing the hang | 22:13 |
| cjwatson | the latter has a trace of partition operations | 22:14 |
| cjwatson | regarding "more convenient", what I strongly recommend for manual partition additions is to use labels | 22:14 |
| cjwatson | they're easier to remember than partition numbers and more robust, at the small cost of having to remember to set the label once per filesystem creation | 22:15 |
| em22_ | yes, the behavior is strange.... my observation is that everything was working perfectly, and I did multiple installs of 12.10 and multiple aborted installs of 13.04 with out any problems. but as soon as I selected to delete a partiton followed by reverting that deletion. after that I can no longer install to that disk. so we must then wonder what is the potential cause of such behavior. furthermore, this change is persistant | 22:15 |
| em22_ | /permenant. Even after powering off the computer, even after using a DVD instead of the flash drive. I have a disk that used to work fine for installs, and now it no longer works for installs. therefore something has been permanently altered, so the question becomes what? | 22:15 |
| cjwatson | I'm not asserting that it's unrelated, merely saying that it's surprising and my instinct as a developer would be to look somewhere else first | 22:16 |
| cjwatson | anyhow, bed | 22:16 |
| em22_ | cjwatson: ah, yes, well, one of the annoyances of the installer's partition manager is that it does NOT give any ability to specify a label for your disk, so be default you end up with a bunch of unlabed volumes. | 22:17 |
| cjwatson | yeah, I have an old bug about that. but it's not a problem for this case since any partitions you set up in the installer's partition manager will be configured using UUIDs for you | 22:17 |
| cjwatson | (assuming you opt to mount them somewhere, of course, and why not) | 22:17 |
| em22_ | until you try to access them for a different os and then find they have no lable | 22:17 |
| cjwatson | if you're setting up partitions by hand later, you're also not using the installer's partition manager | 22:17 |
| cjwatson | sure, easy enough to set though at the point when you're writing fstab lines by hand | 22:18 |
| ogra_ | so add one if thats your usecase | 22:18 |
| cjwatson | I mean, this is already not in pretty autoconfig UI territory | 22:18 |
| ogra_ | its not like it is hard to add an additional label | 22:18 |
| cjwatson | I mean, I would like to fix that annoyance in my copious free time (ho ho), but it shouldn't block you here. | 22:19 |
| ogra_ | heh | 22:19 |
| em22_ | old habits die hard.... it is easy to use a number, anything else requires jumping through extra hoops, if the install it temporary the hoops may not be woorth the effort. | 22:20 |
| cjwatson | this is an old habit that should have been killed off at the point when Linux started making it very, very obvious that it was not going to guarantee disk device names across kernel versions or even (sometimes) across reboots. | 22:21 |
| em22_ | okay, I concede that most people are not going to be installing 5 oses.... but the general pricaple remains that it is dangerous to delete a partition in the middle, if you do it you had better know what you are doing or you will get in trouble. | 22:21 |
| cjwatson | that's why we put considerable effort into migrating away from plain disk device names in persistent configuration files. | 22:21 |
| cjwatson | by definition the manual partitioning UI is for people who know what they are doing | 22:21 |
| em22_ | okay point taken | 22:22 |
| cjwatson | I'm not completely opposed to discreet warnings if they don't get in the way (so no modal dialog boxes), but we aren't going to start inserting layer upon layer of "no, even more advanced" UIs | 22:22 |
| em22_ | a tool tip over the add remove buttons woud be nice, they are quite obtuse | 22:23 |
| cjwatson | bug would be welcome | 22:23 |
| cjwatson | I should really sleep :) | 22:24 |
| em22_ | okay, where would you like me to file the bug... launchpad? what area etc? | 22:24 |
| em22_ | thankks for the chat, have a good night | 22:24 |
| cjwatson | https://bugs.launchpad.net/ubuntu/+source/ubiquity/+filebug | 22:25 |
| cjwatson | or the ubiquity source package in Ubuntu if approaching from some other direction | 22:26 |
| cjwatson | (the first url may kick you out in favour of 'ubuntu-bug ubiquity' or something, I forget) | 22:26 |
| SilentBot | Hello | 23:17 |
| SilentBot | does anyone in here know why the ubuntu installer will not see my windows 8 partition? | 23:17 |
| SilentBot | Ive followed the guide for installing on a UEFI system but with success | 23:18 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!