Javier Martinez wrote: > El 7/10/25 a las 17:37, Dale escribió: >> Javier Martinez wrote: >>> El 7/10/25 a las 16:04, Dale escribió: >>>> Howdy, >>>> >>>> As most everyone knows, I have some large file systems here. Mostly >>>> videos. These are encrypted until I open/unlock them. So, when I >>>> am on >>>> those rare occasions booting up, there's no way to run fsck on them >>>> automatically. I basically have two questions on this. 1: How often >>>> should I open but not mount the file system and run fsck on it? >>>> Once a >>>> month, two months, six months, a year or what. 2: Also, what is a >>>> command that you use that will fix most things without asking a lot of >>>> questions but not do anything that will damage or potentially damage a >>>> file? Basically, what command would you use for a situation like >>>> this? >>>> The man page shows a option for all 'yes' but there may be times when >>>> 'yes' isn't a good idea. I'm wondering if someone has came up with a >>>> way to handle this with some option I'm not aware of. >>>> >>>> Right now I use the command e2fsck -pf /dev/mapper/<mount point >>>> here> to >>>> check it but it ignores some things that it would usually fix if I >>>> were >>>> being asked first. Something about things could be smaller. It's >>>> usually a LOT of them. I'd like those to be corrected as well. Maybe >>>> running the check twice with different options will fix it all???? >>>> >>>> Current info. >>>> >>>> >>>> FILESYSTEM (=) USED >>>> FREE (-) %USED USED AVAILABLE TOTAL MOUNTED ON >>>> /dev/mapper/crypt >>>> [==========================================--------] 83.8% 39.5T >>>> 7.6T 47.1T /home/dale/Desktop/Crypt >>>> /dev/mapper/data >>>> [==============================================----] 91.7% 43.2T >>>> 3.9T 47.1T /home/dale/Desktop/Data >>>> >>>> >>>> I'm thinking about adding a 18 or 20TB drive to data. I got a >>>> empty one >>>> in my safe as a spare. May need to buy a couple more drives soon. >>>> >>>> Thanks. >>>> >>>> Dale >>>> >>>> :-) :-) >>>> >>> >>> You can make one script involving cryptsetup and fsck. Or you can add >>> one passfile to the luks volume add it to the crypttab and to the >>> fstab to auto unlock at booting and check it. >>> >>> I suggest you that you make the check every time you are going to >>> mount your disk before doing it. If the filesystem has some kind of >>> error could be problematic. If its in healthy state, it will not spend >>> time, >>> >>> Maybe you can use "-E no_optimize_extents" option in e2fsck if the >>> message that is boring you is related with extents optimization. >> >> I had actually thought about including the check in my open script. >> It's the little script I made to open and mount the file system. I >> don't want to use anything that opens the file systems automatically. I >> open and close them manually as needed. >> >> The one I get the most is something like this. >> >> >> Inode 761564625 extent tree (at level 1) could be shorter. Optimize? >> yes >> >> >> Sometimes it says narrower instead of shorter. There are hundreds of >> them but after a few, it gives me a "a" option to say 'yes to all' of >> those types of questions. The thing I don't like about the -p option, >> it seems to only run certain tests but also ignores some things that it >> might be best to correct as well. When I leave off the -p option, it >> takes longer but it does more testing. Tests it doesn't do with the -p >> option. I want it to be thorough in the testing but I also would like >> it to fix what it can without me typing in "y" for each one, or quite a >> few of them before giving me a yes to all option. >> >> I figure someone on this list has ran into this problem and found a >> really good solution. I'm just hoping someone will share. LOL ;-) >> >> Dale >> >> :-) :-) >> > The option no_optimize_extents or even fixes_only should make this > questions go away even in /etc/e2fsck.conf, I would not use -y option, > -n instead to answer all no. Also you can use -z undo_file to allow > restoring if fsck makes one mistake > > I suggest you that you use -ck periodically in case that your harddisk > start having faulty sectors. You can use tune2fs -c flag to force a > check between mounts or use -i to force a complete check between > periods of time > > So, in resume, always make a fsck check before mount to see if its in > a clean state or not. > > Use tune2fs to force a check periodically between mounts or time > passed even if it seems clean. > > If you don't want to take care of extents optimizations, you can set > options sent before in /etc/e2fsck.conf > > I would never use "-y" option in fsck. > > If you want to get fsck into the script you should check fsck error > codes before mounting the volume, so $? is your friend. > > 0 - No errors > 1 - File system errors corrected > 2 - File system errors corrected, system should > be rebooted > 4 - File system errors left uncorrected > 8 - Operational error > 16 - Usage or syntax error > 32 - E2fsck canceled by user request > 128 - Shared library error > > If not 0 or 1 don't try to mount it.
This is one reason I'm thinking about putting it in my script so it does it each time. When I leave home, I always close my encrypted file systems. When I get back home, I open them again. I usually go to town once a week, standing Doctor office visit and gotta have groceries. So, I do unmount once a week at least. Thing is, a lot can go on during that week. Some weeks the amount of data that changes can be fairly large. Some weeks, not much at all. I just want to figure out if it is something I should just do manually every once in a while or include in my script, just to be sure. I read up on the -y option a while back. I don't plan to use that. >From what I've read, most of the time it is fine but it can think it is fixing something when it is actually not. I've read where some in certain cases lost data using that. I don't mind it making things shorter or narrower. Those seem to be normal fixes. Thing is, with the -p option, it ignores that and seems to not run all the tests. Without -p it seems to do a lot more. I figure I need a better set of options to get the check done right without it asking me a lot. That way, I can put it in a script as well. I read the man page and the -ck options may require some more reading. Most of the time, man pages don't make sense to me. I have to find better explanations that make more sense. It sounds like it does more checking and updating of bad blocks. That's as far as I got so far but sounds like a good idea. I'm familiar with $?. I'm not sure how to make it check for two outputs tho. Right now, want to find out the manual way. Then move it to a script. Making progress. Dale :-) :-)
