On Fri, 12 Aug 2022 at 03:17, Steve French <smfrench@xxxxxxxxx> wrote: > > The "jiffies vs. seconds" in comment was the only suggestion I didn't include. > See updated patch v2 (attached), I made minor updates. Added the > Suggested-by from Bharath. Moved the defines for default/max to > different name with SMB3 (and in fs_context.h) since it is an smb3 > feature (so not confused with cifs). I increased the default to 5 > seconds (although that is still lower than some other clients - it > should help perf. As you suggested, unconditionally print the value > used on the mount. > for some workloads). nack. The problem with this is that it is a mount option that is impossible for a sys admin to set correctly. If we need this as a mount option we need documentation on it too. 1, How does a sys admin determine that there is an issue and that changing this value will fix it? 2, How does a sys admin determine what to set it to? To me it seems this is an option that can only be used by developers and thus it should not be a mount option. We have too many ad-hoc mount options that end users can not use correctly as it is. > > On Thu, Aug 11, 2022 at 11:16 AM Paulo Alcantara <pc@xxxxxx> wrote: > > > > Steve French <smfrench@xxxxxxxxx> writes: > > > > > Will fix the typos thanks. > > > > Thanks. > > > > > There are a couple of minor differences from Bharath's earlier patch e.g. > > > > > > "closetimeo" rather than "dclosetimeo" (I am ok if you prefer the longer name), > > > and also this mount option is printed in list of mount options if set. > > > > Both look good to me. I personally don't care much about naming, > > though. > > > > -- > Thanks, > > Steve
