cryptic error message when TPC not supported

[cshimmin]

This is perhaps related to #270, but I found that when running xrdcp --tpc only ... with a destination that does not support TPC, I get the following rather unhelpful error message:

Run: [ERROR] Server responded with an error: [3011] Unable to write file; r/o file already exists.

It took me quite some time to find out that I can query the destination with query config TPC, only to learn that the destination doesn't support TPC.

I understand from #270 it's not possible to determine beforehand whether TPC is supported on the destination. But is there any reason one couldn't print a more helpful message after, e.g. something like The destination site does not support TPC.?

[abh3]

This message has nothing to do with TPC support. The message should be taken literally. It means that you are trying to write a file (i.e. open in r/w mode) and could potentially do so if it were not for the fact that some node in the cluster is exporting that very same file in r/o mode. In this case, you cannot open the file r/w.

[cshimmin]

Sorry, I'm still a bit confused. Before running xrdcp --tpc only -p <some_source_file> <some_dest_file>, the file does not exist at all on the destination cluster at all. So how is it that a node can be already exporting it in r/o mode?

[cshimmin]

I noticed now that after running the above command, the file does exist, but is 0 bytes and is listed with r/w permissions (when I log in). But I am guessing that the source site does not have write permissions on this file, so that is consistent with what you were saying.

So I think I understand now your point, but it's still a confusing message for a new user ;)

[abh3]

That I cannot tell you. However, you can always find out the state of the
file by using xrdfs : locate -n

this will tell where the file is and accessibility.

Andy

On Wed, 25 Nov 2015, Chase wrote:

Sorry, I'm still a bit confused. Before running xrdcp --tpc only -p <some_source_file> <some_dest_file>, the file does not exist at all on the destination cluster at all. So how is it that a node can be already exporting it in r/o mode?

---

Reply to this email directly or view it on GitHub:
#308 (comment)