Fedora Account System
Red Hat Associate
Red Hat Customer
It appears that the redhat-config-* apps will be the moral equiv of AIX's SMIT/SMITTY, HP-UX's SAM, and Solaris's SMC. One VERY nice feature of AIX's SMIT/SMITTY is that you can elect to see EXACTLY what commands and changes are being made "behind the scenes". Please consider adding such functionality to the redhat-config-* apps. Maybe a "show work console window" button or logging all everything to a file in /var/log/. Digging through the source and/or running strace/ltrace on the apps to figure out exactly what they are doing is not fun. Another benefit is that the legions of former Windows sys admins who will be using the redhat-config-* tools will have an easier time becoming 'command line jockeys'. A major problem with all the Windows admin tools is that they 'hide' what's going on. The whole "hood welded shut" phenomenon. Please keep the redhat-config-* tools from having their hoods welded shut.
Thanks for adding the RFE -- The development language aspect to some degree is driving how much this can be done -- within python, one may either open a file handle for logging to, or perhaps a preferred method would be to describe a 'sufficient' summary of the net config file changes with the syslog. By using a scheme like: echo "echo "rh-c-n eth1 static 192.168.1.2 255.255.255.0." | \ /usr/bin/logger -p local0.info once gets a truly portable way to record the information. It may be beter to use a series of logging entries: echo "echo "rh-c-n device=eth1 " | \ /usr/bin/logger -p auth.warning echo "echo "rh-c-n proto=static " | \ /usr/bin/logger -p auth.warning and so forth ... the transaction series will automatically get the process ID added by the syslog, so we can differientiate transaction groups. What a concept -- using small *nix tools to build better tools.
The question of what, and how to log at initial install/upgrade Anaconda time remains -- currently we append into the file down in /tmp -- having a syslog thread running early is possible, but it is not possible to start it _too_ early in an install or upgrade, because there may be no partition yet to hold the destination syslog logging file. A couple of approaches appear -- hold a temp file until the destination is present, or just stay with the logging file, and maybe write a late process step to dump, line by line, the conents of that file to a 'logger' once it _is_ possible to start a syslogd up ...
Created attachment 71325 [details] /var/log/explanations from Mandrake 9.0 beta 3
alexl asked for example Mandrake DrakConf /var/log/explanations. The above attached is from my friend's computer. It would be really nice if Red Hat's tools did something similar, perhaps to the same log filename.
To me, the install/upgrade case is handled by anaconda and uses the files /root/install.log and /root/upgrade.log. Are the redhat-config-* apps even used during install/upgrade? The use of the redhat-config-* apps post install are, to me, the important case.
Thanks, Warren -- the content attached looks just like what my suggestion would have produced with logger, and an entry in the /etc/syslogd.conf file: local0.* /var/log/explanations
Later, perhaps.
Reopen as an RFE for post RHL 8.0 consideration in planning items for next release
http://people.redhat.com/harald/redhat-config-network with -vvvvvv you'll see most writing/modifying actions
As a matter of course (ie, automatically and always), all the redhat-config-* apps should log what they do. To the file /var/log/redhat-config or something. This RFE doesn't apply just to redhat-config-network, but to all the redhat-config apps.
Any progress on this?
redhat/system-config-network writes its changes to /var/log/{redhat,system}-config-network