Results from unique ID 3e1b6b19063c6(209.240.198.62) on 07/01/2003 at 04:16:53 pm: Version: 01/07/2003 - 12:53:50 am -------------------------------------------------- *** Environment *** Protocols: , , SMTP, , , , Products: postfix Platform: Intel Linux *** Experience *** Admin Length: v5 Admin Responsibility: v5 Responsibilities: v1 Environment Length: v4 Certifications: *** User Base *** Number of Users: v1 Volume: v1 Regular Downtime: v4 Off-Hours?: no Weekday Off-Hours: N/A Weekend Off-Hours: N/A Off-Hours Downtime: v7 Unplanned Downtime: v4 Percentage of Complaints: v6 Number of Problems: v1 *** Tasks *** User Requests: v5 Diagnostics vs. Repairs: v2 Common Task 1: mail not being delivered Common Task 2: Common Task 3: Lost E-mail: Lost E-mail Cause: administrative, log files filled up file system Challenging Tasks: Confidence Time: v4 *** Further Discussion *** Other Comments: mail administration can be deceptive problem. It often appears simple but in reality is much more complex. It is an application that fortunately works \"well enough\" in the vast majority of the cases. This makes it really easy to reach a incorrect assumption about the what is needed for a working system. Often administrators don\'t understand enough of the underlying issues. Often the protocols or systems underlying mail delivery have behaviours that are rarely seen under normal situations. Expanded Comments: Results from unique ID 3e1b6a9973a5a(12.233.34.29) on 09/01/2003 at 04:20:23 am: Version: 01/07/2003 - 12:53:50 am -------------------------------------------------- *** Environment *** Protocols: IMAP, POP3, SMTP, , , , Products: Exim (SMTP), UW ipopd-ssl, UW imapd-ssl Platform: Debian GNU/Linux (kernel 2.4.19), 1xAMD Athlon 750MHz CPU, 256MB SDRAM, 512MB swap, 10/100Mbps NIC (Linux tulip driver) *** Experience *** Admin Length: v3 Admin Responsibility: v3 Responsibilities: v2 Environment Length: v2 Certifications: *** User Base *** Number of Users: v3 Volume: v2 Regular Downtime: v4 Off-Hours?: no Weekday Off-Hours: N/A Weekend Off-Hours: N/A Off-Hours Downtime: v1 Unplanned Downtime: v3 Percentage of Complaints: v2 Number of Problems: v3 *** Tasks *** User Requests: v3 Diagnostics vs. Repairs: v5 Common Task 1: Mailing list/mail alias administration; usually driven by direct user request. (approx. once a week, approx. 5 minutes per request) Common Task 2: Clearing frozen messages from Exim\'s mail queue This task typically has no user-observable effects. (approx. once every two weeks, with spikes of several times in a day; approx. 5 minutes per occasion) Common Task 3: Lost E-mail: Due to network changes, direct email service was unavailable for a period of several hours. Incoming mail was queued on the backup MX (which did not suffer from these problems), but users were unable to retrieve mail or send mail through our server. This outage caused a few (2-3) user complaints. Lost E-mail Cause: A combination of external problems and adminstrative inaction caused the problem. Challenging Tasks: Altered Exim\'s handling of undeliverable messages. If an inbound message cannot be delivered, Exim bounces it; if the bounce cannot be delivered, Exim \"freezes\" the original message and alerts the administrator. The message remains in Exim\'s queue for a week; it is then retried, and if delivery fails again, the message is discarded. These notifications are annoying, since no action is possible; furthermore, the messages are, so far as I\'ve noticed, entirely spam. Exim\'s configuration was modified to not inform the administrator of these events. The appropriate option was difficult to discover in Exim\'s documentation. Confidence Time: v2 *** Further Discussion *** Other Comments: The learning curve for many administration tasks is decidedly non-linear. While I may be able to diagnose and correct most common problems that I see, I can only do this after a fairly long period of familiarization with the system. Most of the time, the mail server requires no intervention on the part of the administrator after initial setup; during the brief (and usually unwelcome) times of activity, familiarity with the particular system in use is the most important tool for problem resolution. This is particularly true when dealing with email systems, as email is an important communication channel which cannot be unavailable for extended periods; there is no time for an administrator to check documentation, which is often lacking or poorly laid out, while an organization\'s email capability is reduced or unavailable. Personal experience with the specific system in use is therefore critical; since critical situations do not typically occur daily (one hopes), this experience can be hard to obtain. Expanded Comments: Results from unique ID 3e1daaa6da298(171.64.64.112) on 09/01/2003 at 09:11:19 am: Version: 01/07/2003 - 12:53:50 am -------------------------------------------------- *** Environment *** Protocols: IMAP, POP3, SMTP, Webmail, , , Products: exim, sendmail, UW imapd, Stalker Communigate Pro Platform: sparc/solaris, i386/solaris, i386/linux *** Experience *** Admin Length: v5 Admin Responsibility: v3 Responsibilities: v2 Environment Length: v2 Certifications: *** User Base *** Number of Users: v5 Volume: v5 Regular Downtime: v3 Off-Hours?: no Weekday Off-Hours: N/A Weekend Off-Hours: N/A Off-Hours Downtime: v1 Unplanned Downtime: v2 Percentage of Complaints: v1 Number of Problems: v2 *** Tasks *** User Requests: v3 Diagnostics vs. Repairs: v3 Common Task 1: legitimate email incorrectly tagged as spam (false positives) Common Task 2: spam not tagged (false negatives) Common Task 3: removing a host / sender address from our blacklist Lost E-mail: Lost E-mail Cause: Challenging Tasks: trying to find a solution for forged \'From\' addresses that make it past the spam filter. Since we don\'t want any @stanford.edu emails marked as spam, they are on a whitelist, but this allows spam with a forged stanford address to make it through. We found a solution with a couple of Exim ACL changes and a new feature in Spamassassin that seems to have solved the problem. Confidence Time: v4 *** Further Discussion *** Other Comments: The vast majority of user questions / complaints are not a result of failure of the mail system, but rather are caused by other things like disk quotas, new versions of mail readers that behave differently, and misunderstandings by users (for example, that from off-campus they can\'t use our SMTP servers as relays). Expanded Comments: Results from unique ID 3e1f1ef63e3c8(169.229.3.38) on 10/01/2003 at 11:49:41 am: Version: 01/07/2003 - 12:53:50 am -------------------------------------------------- *** Environment *** Protocols: IMAP, , SMTP, Webmail, , , Products: iplanet and sendmail Platform: imap server Sun-Fire-880 running Solaris 9, various Sun hardware running Solari 8 and 9 for TLS outgoing smtp gateway and other various smtp related services *** Experience *** Admin Length: v5 Admin Responsibility: v5 Responsibilities: v2 Environment Length: v3 Certifications: *** User Base *** Number of Users: v5 Volume: v5 Regular Downtime: v3 Off-Hours?: yes Weekday Off-Hours: Weekend Off-Hours: 8 Off-Hours Downtime: v7 Unplanned Downtime: v2 Percentage of Complaints: v1 Number of Problems: v2 *** Tasks *** User Requests: v3 Diagnostics vs. Repairs: v1 Common Task 1: Scheduling forwarding changes Common Task 2: Monitoring server performance for possible problems Common Task 3: Investigating spam and or virus related problems Lost E-mail: 1) email forwarding database problem due to problems with NFS locking on new file server 2) lost e-mail after imap server upgrade due to unexpected configuration problem 3) degraded service when first deploying an anti-spam flagging software Lost E-mail Cause: Primarily software problems with some non-optimum hardware interactions (NFS locking, hardware not beefy enough for anti-spam software) Challenging Tasks: Deploying SpamAssassin in our non-standard environment (servers with no local user accounts or /var/mail or procmail) provided some challenges Confidence Time: v2 *** Further Discussion *** Other Comments: Expanded Comments: Results from unique ID 3e268346a78f6(68.130.93.11) on 16/01/2003 at 02:03:55 am: Version: 01/16/2003 - 02:00:44 am -------------------------------------------------- *** Environment *** Protocols: IMAP, , SMTP, Webmail, , , Products: testing Platform: testing *** Experience *** Admin Length: v3 Admin Responsibility: v2 Responsibilities: v2 Environment Length: v3 Certifications: testing *** User Base *** Number of Users: v5 Volume: v2 Regular Downtime: v4 Off-Hours?: yes Weekday Off-Hours: 5 Weekend Off-Hours: 14 Off-Hours Downtime: v3 Unplanned Downtime: v1 Percentage of Complaints: v5 Number of Problems: v6 *** Tasks *** User Requests: v6 Diagnostics vs. Repairs: v3 Common Task 1: testing Common Task 2: testing Common Task 3: testing Lost E-mail: testing Lost E-mail Cause: testing Challenging Tasks: testing Confidence Time: v4 *** Further Discussion *** Other Comments: testing Expanded Comments: testing