Re: [Rpm-maint] [rpm-software-management/rpm] Size check is too strict (#1232)
We are not looking deeper into this. Closing as I don't really see anything wrong with rpm's behaviour. -- You are receiving this because you are subscribed to this thread. Reply to this email directly or view it on GitHub: https://github.com/rpm-software-management/rpm/issues/1232#issuecomment-760945858___ Rpm-maint mailing list Rpm-maint@lists.rpm.org http://lists.rpm.org/mailman/listinfo/rpm-maint
Re: [Rpm-maint] [rpm-software-management/rpm] Size check is too strict (#1232)
Closed #1232. -- You are receiving this because you are subscribed to this thread. Reply to this email directly or view it on GitHub: https://github.com/rpm-software-management/rpm/issues/1232#event-4211188833___ Rpm-maint mailing list Rpm-maint@lists.rpm.org http://lists.rpm.org/mailman/listinfo/rpm-maint
Re: [Rpm-maint] [rpm-software-management/rpm] Size check is too strict (#1232)
Double checking the code the unchanged files should not be counted when `%_minimize_writes` is enabled. So it obviously isn't in this test case. The question here is why the installation with `--ignoresize` passes without an error. I wonder if there is some trickery going on on the FS level to not actually using all those blocks filled with zeros. From an rpm POV this should not work and rpm is expected to actually use twice as much space unless `%_minimize_writes` is enabled. Not sure if this is really worth further exploring... -- You are receiving this because you are subscribed to this thread. Reply to this email directly or view it on GitHub: https://github.com/rpm-software-management/rpm/issues/1232#issuecomment-656739157___ Rpm-maint mailing list Rpm-maint@lists.rpm.org http://lists.rpm.org/mailman/listinfo/rpm-maint
Re: [Rpm-maint] [rpm-software-management/rpm] Size check is too strict (#1232)
Yeah, upgrades temporarily needing twice the disk-space is to be expected and not a bug. That `%_minimize_writes` actually affects disk-space consumption estimation is an interesting side-effect that I hadn't even thought of until now, but of course it does. 4.16 will auto-enable this if non-rotational disk(s) is detected, but in 4.15 it'll need to be manually enabled. There's nothing wrong with enabling it on rotational disks if minimizing space consumption is important, it'll just have different performance characteristics. -- You are receiving this because you are subscribed to this thread. Reply to this email directly or view it on GitHub: https://github.com/rpm-software-management/rpm/issues/1232#issuecomment-631258061___ Rpm-maint mailing list Rpm-maint@lists.rpm.org http://lists.rpm.org/mailman/listinfo/rpm-maint
Re: [Rpm-maint] [rpm-software-management/rpm] Size check is too strict (#1232)
Well, normally rpm saves the new file to disk first and then moves it over the old file. So it actually does need this space. The latest version has some optimization for SSDs where files that are unchanged are only checksummed. This may be why you don't see an error message here. -- You are receiving this because you are subscribed to this thread. Reply to this email directly or view it on GitHub: https://github.com/rpm-software-management/rpm/issues/1232#issuecomment-630793756___ Rpm-maint mailing list Rpm-maint@lists.rpm.org http://lists.rpm.org/mailman/listinfo/rpm-maint
[Rpm-maint] [rpm-software-management/rpm] Size check is too strict (#1232)
With VM in OpenStack or in AWS, it happens quite often that the rpm tells me: ``` installing package FOO needs 220MB on the / filesystem ``` It quite often happens to me with `linux-firmware` which is big and needs nearly 300 MB unpacked, which is on 3GB of cloud image a lot. This statement is nearly always false because it is counted with a worse case scenario, where all files in the new package are different. Which is rarely true. Usually, it is quite opposite - very few files changes. In such cases the calculation is incorrect. Here is a reproducer: 1) download packages from http://miroslav.suchy.cz/fedora/rpm-size-reproducer/ Both packages install two files in `/var/tmp/`. Both install `/var/tmp/big-file` which is 830 MB big (made out of zeros, so the compressed rpm is very small). Version one contains /var/tmp/one.txt and version two contains /var/tmp/two.txt. Both are small text files with different content. 2) ``` sudo rpm -Uvh /home/mirek/rpmbuild/RPMS/x86_64/big-file-1-1.x86_64.rpm Verifying... # [100%] Preparing... # [100%] Updating / installing... 1:big-file-1-1 # [100%] ``` 3) Now run something like: ``` dd if=/dev/zero of=/var/tmp/foo1 ``` tune this so you do not consume whole disk space, and leave few hundred of free space. In my case: ``` $ df -h /var/tmp/ Filesystem Size Used Avail Use% Mounted on /dev/mapper/fedora-root 50G 47G 197M 100% / ``` 4) Now try to upgrade: ``` $ sudo rpm -Uvh /home/mirek/rpmbuild/RPMS/x86_64/big-file-2-1.x86_64.rpm Verifying... # [100%] Preparing... # [100%] installing package big-file-2-1.x86_64 needs 674MB on the / filesystem ``` This is obviously incorrect as `/var/tmp/big-file` does not change. Definitely not the size. 5) Let's try to ignore size-check: ``` $ sudo rpm -Uvh /home/mirek/rpmbuild/RPMS/x86_64/big-file-2-1.x86_64.rpm --ignoresize Verifying... # [100%] Preparing... # [100%] Updating / installing... 1:big-file-2-1 # [ 50%] Cleaning up / removing... 2:big-file-1-1 # [100%] ``` Version: `rpm-4.15.1-3.fc32.1.x86_64` -- You are receiving this because you are subscribed to this thread. Reply to this email directly or view it on GitHub: https://github.com/rpm-software-management/rpm/issues/1232___ Rpm-maint mailing list Rpm-maint@lists.rpm.org http://lists.rpm.org/mailman/listinfo/rpm-maint