You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
We're using icommands & server both 4.2.7 on Ubuntu Bionic.
The writing commands we (we locally being the /cgp zone) use are
iput of "small" files: doesn't tell the destination
iput of "big" files: -V flag will tell destination server, but not the resource
irepl, any size: doesn't tell the destination
I'm asking specifically about write commands because we're seeing some odd performance problems which @kript and @bh9 are dealing with for us, but I can't so easily generate stats on "what transfers were fast, and which were slow?" when the icommands don't say where the file was put.
Things that would be useful,
tell the DATA_ID of the resulting object
tell the DATA_RESC_ID(s) of resulting copies as they're created
machine readable output
consistent output format & contents, regardless of the different code paths used to do the work
consistent output format, even when the operation fails
efficiency; using information already to hand, not spending time or generating load by doing more lookups.
I realise this isn't going to happen in time to deal with our current problem, but please can it guide future changes to icommand output?
Without the above I may look at the before & after of irepl happening. One significant advantage of this in the current contrext is that I have "before" already, with the DATA_ID of these files in bulk. From that I can discover later where these files ended up, and it's much more efficient to do them in bulk:
This way I have the option of patching over the lack of verbosity with more programming but less runtime, because I can query around 1000 individual files or a range of 100k files in one bite.
Current outputs look like this, when I ask for Christmas-tree verbosity:
We're using icommands & server both 4.2.7 on Ubuntu Bionic.
The writing commands we (we locally being the
/cgp
zone) use are-V
flag will tell destination server, but not the resourceI'm asking specifically about write commands because we're seeing some odd performance problems which @kript and @bh9 are dealing with for us, but I can't so easily generate stats on "what transfers were fast, and which were slow?" when the icommands don't say where the file was put.
Things that would be useful,
I realise this isn't going to happen in time to deal with our current problem, but please can it guide future changes to icommand output?
Without the above I may look at the before & after of
irepl
happening. One significant advantage of this in the current contrext is that I have "before" already, with theDATA_ID
of these files in bulk. From that I can discover later where these files ended up, and it's much more efficient to do them in bulk:This way I have the option of patching over the lack of verbosity with more programming but less runtime, because I can query around 1000 individual files or a range of 100k files in one bite.
Current outputs look like this, when I ask for Christmas-tree verbosity:
so I had since 3.3.1 regarded the
-v
and-V
flags as producing no useful information, and avoided using them.The text was updated successfully, but these errors were encountered: