On Mon, 8 Jul 2002, Hetz Ben Hamo wrote:
> Anyone got any experience with one (or both) of the programs? any comments to > share? i've been using big brother for a year or more. on the bottom line, using big-brother vs using nothing - using big brother is much better, and since i use it, i have no more 'disk full' errors (disk space errors were my most annoying problem. machines coming down - i could see that without big brother). big brother (the version i use anyway, which is more then a year old) has some annoying problems: 1. it can parse /var/log/* (or /vad/adm/*) log files, and send a notification when they have errors. however, when it spots such an error, it does not 'rememebr' that it notified me about it, and thus keeps sending me notifications until the log files get rotated. this is quite annoying. 2. when there is a problem, it keeps sending notifications about it time after time after time. this is a 'good' annoyance, since it forces me to solve the problem. however, if i'm not at work (and we don't run a 24 hours a day operation), a large number of such messages pile up in my mail box. i use pine for email, running directly on the mail server machine, so i don't have to download those messages anywhere, but it might be more problemtic to people who read problems from remote. 3. there is no 'root cause analyses'. if the network is down, i'll start getting many errors about single services. 4. in order for the installation to work properly, the installation sohuld be done on the local directory of each machine (to avoid big brother from stalling in case of NFS problems). however, the 'hosts' file needs to contain all machines - i just put a copy of it on an NFS-mounted directory - and it does not seem to cause problems (possibly its read only once, possibly even only by the server machine? no idea). 5. configuration is not very flexible. for example, you can tell it to warn you in case load average goes above a given number. however, this is not the way to tackle load averages, since they are problemativ only if they last for some given ammount of time, and only during given times. i keep getting high load averages during backups, for example, so i get warnings about them which i'm not interested in. on the other hand, i don't get warnings if a single process has gone ammock (we used to have usch a problem with some X applications, who'd go into 100% CPU usage, if they went loose). this won't make the load average high (only increase it by 1), but will make the machine much slower for compilations, for example. big brother does not seem to help with that. hope this helps, -- guy "For world domination - press 1, or dial 0, and please hold, for the creator." -- nob o. dy ================================================================= To unsubscribe, send mail to [EMAIL PROTECTED] with the word "unsubscribe" in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
