[Error] segfault: memory not mapped

[SudipSinha]

[Error] segfault: memory not mapped

[piccolbo]

Hi,
It's not clear which code you are trying to run because you have the statisticians example and the example from the docs which is what apparently is causing the segfault. For the statisticians example I get an empty output. To make your example work I had to replace relative paths with absolute paths wherever you name a file. I also added a comma in front of drop line 26 to make it clearer to me but that seems to only remove a harmless warning. I am willing to consider supporting relative paths, but if hadoop dfs doesn't have cd and pwd commands, it seems to me we would be trying to add a feature to the hadoop fs which is not really the responsibility of rmr.
As far as the example from the docs, it works for me. Could you please report rmr2, hadoop and os versions? Segfaults shouldn't occur no matter what you type. Thanks

[SudipSinha]

Hi,
Sorry for not mentioning that the 1st piece of code was also giving the same error as the statisticians' example (same stderr logs for the mappers). Also, I must mention that I'm only getting this error on my VM (machine configurations below).

VM config:

R: v2.15.2
rmr: rmr2 v2.2.0
OS: CentOS 6.4
OS: CDH3u4
RAM: 8GB

I also replaced all relative paths with absolute ones, but continue to get the error.

P.S. I agree with the absolute path logic that you've put forward.

[SudipSinha]

Hi,
I tried your suggestion of absolute paths for my example on a cluster, and it worked fine.

Thank you!

P.S. I still keep getting the segfault error in my VM.

[SudipSinha]

I reinstalled rmr2 v2.2.0 and I'm not getting the error anymore. I'm closing this issue, as the problem has been solved. But according to me it still makes sense to figure out why it occurred in the first place.