LOG message goes to standard error

[jielab]

Hi, Xiaowei:

When i use "bsub -o my.LOG -e my.ERR" to run RVTESTS, i found that the log message goes to the my.ERR file instead of the my.LOG file. Usually, I check whether my.ERR file is empty to determine if there is anything wrong. Now, i have no easy way to determine if my RVTESTS jobs finished successfully, when I run hundreds of jobs. So, can you please make the standard log message goes to the "-o" file?

Also, i mentioned this previously a couple of times. The log file always shows the following two lines:
connect() error: Operation timed out
Retrieve remote version failed, use '--noweb' to skip
However, RVTESTS does not have an "--noweb" option. I think RVTESTS is still using PLINK and this message is generated by PLINK. If so, can users have a way to specify PLINK2 to be used instead?

Thanks!

Jie

[zhanxw]

Hi Jie, For the purpose of convention, log messages go to stderr by default ( https://stackoverflow.com/a/4919110/425758). To check whether the job is successful or not, can you check the program return code instead of checking my.ERR? If the rvtests successfully exit, the return code should be zero. Internally, RVTESTS does not use PLINK/PLINK2. I don't understand why the message displays when there is "--noweb" option. I will check again and get back to you. Thanks. Xiaowei

…

On Tue, May 23, 2017 at 6:16 AM, jiehuang001 ***@***.***> wrote: Hi, Xiaowei: When i use "bsub -o my.LOG -e my.ERR" to run RVTESTS, i found that the log message goes to the my.ERR file instead of the my.LOG file. Usually, I check whether my.ERR file is empty to determine if there is anything wrong. Now, i have no easy way to determine if my RVTESTS jobs finished successfully, when I run hundreds of jobs. So, can you please make the standard log message goes to the "-o" file? Also, i mentioned this previously a couple of times. The log file always shows the following two lines: connect() error: Operation timed out Retrieve remote version failed, use '--noweb' to skip However, RVTESTS does not have an "--noweb" option. I think RVTESTS is still using PLINK and this message is generated by PLINK. If so, can users have a way to specify PLINK2 to be used instead? Thanks! Jie — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#31>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AAJoiEuzDUqKBvCTS0tTTeqSdHnOfaRfks5r8sAhgaJpZM4NjicP> .

[zhanxw]

Hi Jie,

For the "--noweb" option, I checked rvtests in a machine without Internet access.

When there was not "--noweb" option, I can see the message: "connect() error: Operation timed out";
When I added "--noweb" option, I did not see that message anymore.

In your cluster, do you see this message every time?
Can you please describe more about your network environment?
I am puzzled right now.

[jielab]

Hi, Xiaowei:

As long as the error message goes away with the "--noweb" option, I am fine with it. Our network environment is very strict, within a government firewall. That is all I could say.

For your suggestion to check return code. Can you please give some details so that I can always check it to make sure that RVTESTS finished successfully? Is such a return code in my.ERR file? For now, i only see log message such as "Analysis took 1100 seconds", at the end.

Best regards,
Jie

[zhanxw]

In the coming new version, RVTESTS will print "RVTESTS successfully exit" at the end. You will then be able to tell RVTESTS finished successfully from the last line of my.LOG file. Will that work?

How about the --noweb option? Does it solve your problem?

[jielab]

Dear Xiaowei: Thanks! “RVTESTS finished successfully” in the last line of the log file would be great. maybe there is no need to write “exit”, because usually exit means there is something wrong. I now use “--noweb” and it does not print out the warning message any more. When will the new version be released? Did you find out what cause the AF number to go wrong in logistic regression analysis? Best regards, Jie From: zhanxw [mailto:[email protected]] Sent: 2017年6月9日 16:43 To: zhanxw/rvtests <[email protected]> Cc: jiehuang001 <[email protected]>; Author <[email protected]> Subject: Re: [zhanxw/rvtests] LOG message goes to standard error (#31) In the coming new version, RVTESTS will print "RVTESTS successfully exit" at the end. You will then be able to tell RVTESTS finished successfully from the last line of my.LOG file. Will that work? How about the --noweb option? Does it solve your problem? — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <#31 (comment)> , or mute the thread <https://github.com/notifications/unsubscribe-auth/AZsvfzB8jBSEcxd07jYTnN0OnfUhh3Ucks5sCa5kgaJpZM4NjicP> . <https://github.com/notifications/beacon/AZsvf8weON4WxbMq4BbZxYYOmZcOW5KFks5sCa5kgaJpZM4NjicP.gif>

[zhanxw]

@jiehuang001 the interim release version is here: http://zhanxw.com/rvtests/experimental/rvtests-20170613-01e018-linux64-static.tar.gz

It prints “RVTESTS finished successfully” at the end of the stdout.

I am working on the AF problem now. It's hard to understand that without some examples though.

[zhanxw]

Now I have changed the default frequency calculation method in this version (https://github.com/zhanxw/rvtests/releases/tag/v2.0.6). The AF should be positive all the time now. Thanks.