It's looks like portage do MD5SUM twice on package sources when doing emerge --fetch. I noted it when fetching some large packages such as kernel on my Gentoo in portage harddisk (USB1) with 64MB of RAM computer box. I wait lone time after message: >>> emerge (1 of 1) sys-kernel/vanilla-sources-2.4.24 to / for good MD5SUM ('MD5 ;-)') message come. >>> Previously fetched file: linux-2.4.24.tar.bz2 MD5 ;-) When this line show I think it do MD5SUM once and again. I need to wait a long time. I think we can't notice this if fetch on computer that have much of RAM because it cached. >>> md5 src_uri ;-) linux-2.4.24.tar.bz2 It not looks like a serious bug. But it affected performance on less RAM machine. I think second MD5SUM is only for ebuild unpacking. It's not neccessary for when fetch. Sorry if portage already do MD5 once. I don't sure about this. Also happened on lastest portage. Reproducible: Always Steps to Reproduce: 1.emerge --fetch some large packages on slow harddisk and less RAM 2.notice the MD5SUM messages and MD5SUM time Actual Results: It's looks like portage do MD5SUM twice - First time after fetched files to check for valid archive. - Second time for validate integrity of archive before unpack (not neccessary). Expected Results: MD5SUM once.
Addition: Only happened when fetch old files !!!
The check after fetch was implemented on user request some time ago, so that `emerge -f` already detects errors and a later unattended compile won't get interrupted by md5 errors. I'll look if we can change it so that `emerge foo` will only do one md5 check (but if it requires major changes you'll have to live with it).
Hmm, I should check the code before I talk: It's already so that the md5 check on fetching is only done if you use --fetchonly. Which portage version are you using and what was the exact command you used ?
It's also check MD5 twice with already fetch file. I usually use 'emerge -f packages...' or 'emerge -uf packages...' on my portable HDD (usb1) which very slow. Using portage 2.0.49-r21 (lastest I have). Hmm. I should try 'emerge package' and 'emerge -f package' and compare MD5 output (same package).
still an issue ?
I haven't seen this in quite some time, closing. Reopen (again) if it pops up :)
