When I mount a SMB share with cifs, I get slow throughput over gigabit ethernet, comparable to what I've seen with systems that do not have the TCP_NODELAY set. Some say that that option should be on by default. But my complaint is that it's not even an option. First, mount.cifs doesn't even support the sockopt option. Second, using "mount -t cifs" with "-o sockopt=TCP_NODELAY" has no impact on throughput (read or write). Reproducible: Always
ok, please only reopen if you can explain why this is a problem of the package not of mount.cifs itself (which should then be reported upstream).
It IS a problem with mount.cifs itself. I would push it upstream if I knew how. (Unless "RESOLVED UPSTREAM" means that you have already done this.)
hmm, not really, no. The easiest way is probably the samba mailinglist (but the one for technical/development things).
I'll see if I can figure anything out. But I'm not confident that I (as a nobody) would have any effect. I had the impression that the distros had some sway in these matters and had fairly direct lines to the maintainers of the packages they distribute. (If you don't, that makes it hard to ensure that all of your stuff is solid and up-to-date.) I'm assuming that since you actively distribute the package that you have the email address of the lead developer. Perhaps it would be more effective if you were to privately email me his email address so I could directly contact him on behalf of the distro. In general, I would not recommend just shrugging off things like this. By distributing the package at all, you're representing it, to some extent. Just because something isn't your fault (because you didn't write the package) doesn't mean it's not your responsibility (as a distro maintainer). I recently read an article that pointed out that this attitude is one of the things holding back FOSS penetration into the market. People say "it's not my problem" and bit-rot ensues. Since I reported the bug, and I'm so hot and bothered about it (not really, but whatever), I'll take responsibility. But I'm going to need some assistance and information from someone such as yourself who knows a hell of a lot more about it than I do. Thanks.
(In reply to comment #4) > I'll see if I can figure anything out. But I'm not confident that I (as a > nobody) would have any effect. I had the impression that the distros had some > sway in these matters and had fairly direct lines to the maintainers of the > packages they distribute. (If you don't, that makes it hard to ensure that all > of your stuff is solid and up-to-date.) I'm assuming that since you actively > distribute the package that you have the email address of the lead developer. > Perhaps it would be more effective if you were to privately email me his email > address so I could directly contact him on behalf of the distro. It is true that being the distro-maintainer gives some additional credibility, but in general it doesn't help much. Especially in the samba project the developers are sometimes reluctant to accept patches concerning the behaviour in some special cases (like being able to explicitly disable capability support). I also don't have any private mail address from a Samba developer nor do I know one personally (at least, not as far as I know :-) > > In general, I would not recommend just shrugging off things like this. By > distributing the package at all, you're representing it, to some extent. Just > because something isn't your fault (because you didn't write the package) > doesn't mean it's not your responsibility (as a distro maintainer). I recently > read an article that pointed out that this attitude is one of the things > holding back FOSS penetration into the market. People say "it's not my > problem" and bit-rot ensues. > > Since I reported the bug, and I'm so hot and bothered about it (not really, but > whatever), I'll take responsibility. But I'm going to need some assistance and > information from someone such as yourself who knows a hell of a lot more about > it than I do. > Well, my problem with such requests is that I have no time for it. Not even a second, sorry. I maintain 100+ packages in Gentoo, I just organized a booth at a fair here in Switzerland, I try to push a new EAPI forward, would have to write summaries for the council, etc. etc. and that's just for Gentoo where I don't get paid for. Besides having to study and to earn money for a living. So, I'm deeply sorry, but I am not able to help you and not because I don't want to but because I don't have time to. The only way to change this situation is by becoming a dev yourself, help recruiting new devs so we can distribute the work load (I'm currently alone in samba) or make a company pay me for my Gentoo work so I can quit my other job.
I'm going to be sneaky and report it in Ubuntu's bug database. If they push it upstream, you'll eventually get the update for free. :)
dropped