hi，how about OpenStack predict？

[2017dadi]

hi， your hdfs__data is very perfect,how about openstack?Because of the paper,there are two parts of deeplog,could you help me?Thanks

[amineebenamor]

You should parse all the OpenStack logs, make session windows based on PID and train the LSTM model on normal logs. Afterwards, you can try and check if you can detect abnormal logs as the dataset is labelled. It's a bit of work but it seems feasible by yourself.

[bolzzzz]

You should parse all the OpenStack logs, make session windows based on PID and train the LSTM model on normal logs. Afterwards, you can try and check if you can detect abnormal logs as the dataset is labelled. It's a bit of work but it seems feasible by yourself.

Hi. I also want to try openstack data. According to the paper the labels for openstack data are based on instance_id. However, most of the raw logs from https://www.cs.utah.edu/~mind/papers/deeplog_misc.html don't contain instance_id. Do you have any idea? Thanks.

[HelenGuohx]

Hi , I met the same problem when parsing OpenStack dataset. Besides, even though I made session windows on logs with instance id, the abnormal sequences are similar to most of the normal sequences. How do you solve the problem? @bolzzzz

[TianT1209]

Hi , I met the same problem when parsing OpenStack dataset. Besides, even though I made session windows on logs with instance id, the abnormal sequences are similar to most of the normal sequences. How do you solve the problem? @bolzzzz

Have you deal with this problem? Thanks a lot