Re: [xmail] POP3 UIDL PROBLEM
Oversea Site via xmail wrote on 06/04/2015 04:25 PM: I have over 1 EMAILS but when I test UIDL, Sometimes, it's not sorting by ID. it's a file system -rw-rw-r-- 1 root root30558 Jun 1 04:46 1433105179992.70d12960.34c.3f8be2.bravo -rw-rw-r-- 1 root root 989 Jun 1 05:04 1433106257177.5ace6960.34c.3fafbd..bravo -rw-rw-r-- 1 root root 162486 Jun 1 13:04 1433135071576.71513960.34c.43c7a9.bravo -rw-rw-r-- 1 root root21401 Jun 1 13:47 1433137677235.6fd10960.34c.4428cc.bravo -rw-rw-r-- 1 root root54084 Jun 1 22:24 1433168674933.6ed0e960.34c.48d596.bravo -rw-rw-r-- 1 root root 9112 Jun 2 02:35 1433183738295.6ed0e960.34c.4acc30..bravo -rw-rw-r-- 1 root root 9142 Jun 2 02:35 1433183738434.5f4ef960.34c.4acc39.bravo -rw-rw-r-- 1 root root 5342 Jun 2 03:26 1433186769695.70511960.34c.4b30b7.bravo -rw-rw-r-- 1 root root 1238 Jun 2 04:02 1433188925598.6dd0c960.34c.4b7832.bravo -rw-rw-r-- 1 root root 1020 Jun 2 04:02 1433188949218..69d04960.34c.4b78f7.bravo -rw-rw-r-- 1 root root22474 Jun 2 06:32 1433197934458.6f50f960.34c.4ca2e8.bravo -rw-rw-r-- 1 root root 2615 Jun 2 07:40 1433202046360.59ce4960.34c.4d2b2f.bravo -rw-rw-r-- 1 root root 2678 Jun 2 07:43 1433202184456.4fcd0960.34c.4d2fbe.bravo -rw-rw-r-- 1 root root 3272 Jun 2 08:04 1433203475828.55cdc960.34c.4d5ab8.bravo -rw-rw-r-- 1 root root 2863 Jun 2 08:26 1433204809317.57ce0960.34c.4d8766.bravo -rw-rw-r-- 1 root root41748 Jun 2 11:30 1433215840379.72d16960.34c.4f3bf6.bravo -rw-rw-r-- 1 root root 1857 Jun 2 11:39 1433216376230.73d18960.34c..4f5305.bravo -rw-rw-r-- 1 root root 6157 Jun 2 11:44 1433216684073.71d14960.34c.4f5e9a.bravo -rw-rw-r-- 1 root root 5402 Jun 2 12:55 1433220946062.6f50f960.34c.500170.bravo -rw-rw-r-- 1 root root29790 Jun 2 14:03 1433224988810.70511960.34c.5092ca.bravo -rw-rw-r-- 1 root root10473 Jun 2 22:11 1433254291563.4a4c5960.34c.553526.bravo -rw-rw-r-- 1 root root 1238 Jun 3 04:02 1433275327112.6f50f960.34c.580f59.bravo -rw-rw-r-- 1 root root 1020 Jun 3 04:02 1433275353273.654fb960.34c.581041.bravo -rw-rw-r-- 1 root root 988 Jun 3 05:14 1433279643381.5ecee960.34c.589f69.bravo -rw-rw-r-- 1 root root15235 Jun 3 07:50 1433289052265.614f3960.34c.59db64.bravo -rw-rw-r-- 1 root root 3985615 Jun 3 08:36 1433291762335.49cc4960.34c.5a40b9.bravo -rw-rw-r-- 1 root root30149 Jun 3 14:52 1433314331266.71513960.34c.5dab73.bravo -rw-rw-r-- 1 root root 5382 Jun 3 14:54 1433314498740.6dd0c960.34c.5db2ce.bravo -rw-rw-r-- 1 root root44713 Jun 3 16:55 1433321724856.74d1a960.34c.5ed68c.bravo -rw-rw-r-- 1 root root 110709 Jun 3 17:54 1433325240897.6dd0c960.34c.5f65fa.bravo -rw-rw-r-- 1 root root 322169 Jun 3 18:56 143332954.6e50d960.34c.5fff22.bravo -rw-rw-r-- 1 root root14174 Jun 3 20:57 146272368.6e50d960.34c.610ed5.bravo -rw-rw-r-- 1 root root27877 Jun 4 00:03 1433347390242.7fc51afdd700.bca.152b.bravo -rw-rw-r-- 1 root root 8116 Jun 4 03:06 1433358408984.5c450960.24ad.336b.bravo -rw-rw-r-- 1 root root 9831 Jun 4 14:05 1433397936295.7347e960.24ad.635a7.bravo -rw-rw-r-- 1 root root90604 Jun 4 19:59 1433399979747.5ec55960.24ad.6860d.bravo -rw-rw-r-- 1 root root 2193650 Jun 4 20:00 1433412426056.43c1f960.24ad.8919a.bravo -rw-rw-r-- 1 root root 2783 Jun 4 20:07 1433419665957.6ead6960.649d.6e32.bravo -rw-rw-r-- 1 root root 4651 Jun 4 20:23 1433420596561.73ae0960.649d.90c8.bravo 13873 1433105179992.70d12960.34c.3f8be2.bravo 13874 1433106257177.5ace6960.34c.3fafbd.bravo 13875 1433135071576.71513960.34c.43c7a9.bravo 13876 1433137677235.6fd10960.34c.4428cc.bravo 13877 1433168674933.6ed0e960.34c..48d596.bravo 13878 1433183738295.6ed0e960.34c.4acc30.bravo 13879 1433183738434.5f4ef960.34c.4acc39.bravo 13880 1433186769695.70511960.34c.4b30b7.bravo 13881 1433188925598.6dd0c960.34c.4b7832.bravo 13882 1433188949218.69d04960.34c.4b78f7.bravo 13883 1433197934458.6f50f960.34c.4ca2e8.bravo 13884 1433202046360.59ce4960.34c.4d2b2f.bravo 13885 1433202184456.4fcd0960.34c.4d2fbe.bravo 13886 1433203475828.55cdc960.34c.4d5ab8.bravo 13887 1433204809317.57ce0960.34c.4d8766.bravo 13888 1433215840379.72d16960.34c.4f3bf6.bravo 13889 1433216376230.73d18960.34c.4f5305.bravo 13890 1433216684073.71d14960.34c.4f5e9a.bravo 13891 1433220946062.6f50f960.34c.500170.bravo 13892 1433224988810.70511960.34c.5092ca.bravo 13893 1433254291563.4a4c5960.34c.553526.bravo 13894 1433275327112.6f50f960.34c.580f59.bravo 13895 1433275353273.654fb960.34c.581041.bravo 13896 1433279643381.5ecee960.34c.589f69.bravo 13897 1433289052265.614f3960.34c.59db64.bravo 13898 1433291762335.49cc4960.34c.5a40b9.bravo 13899 1433314331266.71513960.34c.5dab73.bravo 13900 1433314498740.6dd0c960.34c.5db2ce.bravo 13901 1433321724856.74d1a960.34c.5ed68c.bravo 13902 1433325240897.6dd0c960.34c.5f65fa.bravo 13903 143332954.6e50d960.34c.5fff22.bravo 13904 146272368.6e50d960.34c.610ed5.bravo 13905 143
Re: [xmail] new maintainer for the 1.27 branch
bugs--- via xmail wrote on 06/02/2015 01:31 AM: Hi List, I already made some fixes. See http://www.econtrario.ch/tools/. Maybe you can use them. Yes, it's already on the todo list. Thx. -- cu Uenal ___ xmail mailing list xmail@xmailserver.org http://xmailserver.org/mailman/listinfo/xmail
Re: [xmail] BUG : XMAIL 1.27 with SSL
U.Mutlu via xmail wrote on 05/26/2015 04:38 PM: Bart Mortelmans via xmail wrote on 05/26/2015 03:55 PM: It might be difficult to determine if the problem is with XMail or with the client. And even if for example Tencent actually closes the connection incorrectly, it should be better if XMail would be able to handle such an improperly closed connection. But I'm not sure if you'll find somebody in this forum that will be able to fix this problem. Maybe you could try to get K9 developers involved in this? That seems to be managed much more actively than XMail. With a bit of luck, they can at least find out what they are doing differently compared to for example Outlook and that could cause this problem. Regarding FIN_WAIT2: I suspect it is caused by a firewall issue where some FIN's or ACK's are blocked or wrongly filtered out, or the client side doesn't properly close the socket, or some of the sockets it allocates. One really cannot trust any interpreted scripting language many mobiles do use. Using iptables one can diagnose this by logging the FIN's, ACK's. cu Uenal Oversea Site via xmail schreef op 26/05/15 om 15:09: Now I am using original XMAIL-1.27.tar.gz source without simultaneous POP3 logins, no STUNNEL, use NATIVE XMAIL SSL. The problem still exists, K9-MAIL will report IllegalFormatConversonException when I receive more then 100-200 emails. So It should be XMAIL SSL problem. Because When I reuse STUNNEL to do a POP3 tunnel again, the problem will fix. One more thought: If an application throws an exception without catching and handling it, like is the case above, then this means that the application is badly programmed. A good application would catch the exception and do what is necessary in that state of of the program. In this case one can simply say the app or part of it just badly crashed; no app should behave so. Exceptions can and must be handled by the app. cu Uenal ___ xmail mailing list xmail@xmailserver.org http://xmailserver.org/mailman/listinfo/xmail
Re: [xmail] BUG : XMAIL 1.27 with SSL
Bart Mortelmans via xmail wrote on 05/26/2015 03:55 PM: It might be difficult to determine if the problem is with XMail or with the client. And even if for example Tencent actually closes the connection incorrectly, it should be better if XMail would be able to handle such an improperly closed connection. But I'm not sure if you'll find somebody in this forum that will be able to fix this problem. Maybe you could try to get K9 developers involved in this? That seems to be managed much more actively than XMail. With a bit of luck, they can at least find out what they are doing differently compared to for example Outlook and that could cause this problem. Regarding FIN_WAIT2: I suspect it is caused by a firewall issue where some FIN's or ACK's are blocked or wrongly filtered out, or the client side doesn't properly close the socket, or some of the sockets it allocates. One really cannot trust any interpreted scripting language many mobiles do use. Using iptables one can diagnose this by logging the FIN's, ACK's. cu Uenal Oversea Site via xmail schreef op 26/05/15 om 15:09: Now I am using original XMAIL-1.27.tar.gz source without simultaneous POP3 logins, no STUNNEL, use NATIVE XMAIL SSL. The problem still exists, K9-MAIL will report IllegalFormatConversonException when I receive more then 100-200 emails. So It should be XMAIL SSL problem. Because When I reuse STUNNEL to do a POP3 tunnel again, the problem will fix. ___ xmail mailing list xmail@xmailserver.org http://xmailserver.org/mailman/listinfo/xmail
Re: [xmail] BUG : XMAIL 1.27 with SSL
Oversea Site via xmail wrote on 05/26/2015 03:09 PM: Now I am using original XMAIL-1.27.tar.gz source without simultaneous POP3 logins, no STUNNEL, use NATIVE XMAIL SSL. The problem still exists, K9-MAIL will report IllegalFormatConversonException when I receive more then 100-200 emails. So It should be XMAIL SSL problem. Because When I reuse STUNNEL to do a POP3 tunnel again, the problem will fix. But since you said that no error happens with normal mail-clients (Outlook etc.), then the error must lie in the mobile clients you listed. They seem to not handle properly the closing of the sockets they create. cu Uenal Original Message From: Bart Mortelmans To: t...@os.st, "XMail Users Mailing List" Sent: Tue, May 26, 2015, 3:13 PM Subject: Re: [xmail] BUG : XMAIL 1.27 with SSL Is there any way for you to test if the problem exists if you don't allow simultaneous POP3 logins? Oversea Site via xmail schreef op 26/05/15 om 08:14: POP3Utils.cpp - ADD POP3 Concurrent Login. I used this function over 5 years.. I don't know C++ programming. Sorry :) } if (pPOPMD->ulFlags & POPF_MSG_DELETED) { UPopSendErrorResponse(hBSock, ERR_MSG_DELETED, pPOPSD->iTimeout); ErrSetErrorCode(ERR_MSG_DELETED); return ERR_MSG_DELETED; } UsrGetMailboxPath(pPOPSD->pUI, szMsgFilePath, sizeof(szMsgFilePath), 1); StrNCat(szMsgFilePath, pPOPMD->szMsgName, sizeof(szMsgFilePath)); /* * ADDED CODE START */ FILE *pMsgFile = fopen(szMsgFilePath, "rb"); if (pMsgFile == NULL) { UPopSendErrorResponse(hBSock, ERR_FILE_OPEN, pPOPSD->iTimeout); ErrSetErrorCode(ERR_FILE_OPEN); return ERR_FILE_OPEN; } fclose(pMsgFile); /* * ADDED CODE END */ SysSNPrintf(szResponse, sizeof(szResponse) - 1, "+OK " SYS_OFFT_FMT " bytes", pPOPMD->llMsgSize); if (BSckSendString(hBSock, szResponse, pPOPSD->iTimeout) < 0) return ErrGetErrorCode(); if (pPOPMD->llMsgSize > 0 && Original Message From: Oversea Site via xmail To: "XMail Users Mailing List" Sent: Tue, May 26, 2015, 1:56 PM Subject: Re: [xmail] BUG : XMAIL 1.27 with SSL FEDORA 4 X86_64 GLIBC 2.3.6-3 (I can't upgrade it to the lastest version.. Because I have a chiliasp 3.6.2 32-bit run ASP VBSCRIPT on APACHE 1.3 32-bit software, it need GLIBC 2.3.x and Kernel 2.6) KERNEL 3.3 MOD FROM FC15 (It will display KERNEL 2.6.43 Number). Kernel is not a main problem. Because I have been test 2.6.32 or other. same problem was found. REAL HARDWARE -> INTEL B85M i3-4150 16G RAM SATA3 RAID 1 LINUX SOFTRAID1 MDADM Corp. Email System is not heavily use when I test it in HK TIME midnight.. Anyway I will test this issue in other linux distribution in VM, such as CentOS. (You can give me any linux distribution, let's build the same platform in VM. I use virtualbox. [root@bravo logs]# uname -a Linux bravo 2.6.43-gcc402 #3 SMP Sun May 17 23:10:29 HKT 2015 x86_64 x86_64 x86_64 GNU/Linux [root@bravo logs]# [root@bravo logs]# ldd /var/MailRoot/bin/XMail127v3 linux-vdso.so.1 => (0x77f0e000) libssl.so.0.9.8 => /usr/local/ssl/lib/libssl.so.0.9.8 (0x7f43b00bc000) libcrypto.so.0.9.8 => /usr/local/ssl/lib/libcrypto.so.0.9.8 (0x7f43afd35000) libdl.so.2 => /lib64/libdl.so.2 (0x003e1820) libpthread.so.0 => /lib64/libpthread.so.0 (0x003e1860) libstdc++.so.6 => /usr/lib64/libstdc++.so.6 (0x7f43afa37000) libm.so.6 => /lib64/libm.so.6 (0x003e1800) libgcc_s.so.1 => /lib64/libgcc_s.so.1 (0x003e1940) libc.so.6 => /lib64/libc.so.6 (0x003e17d0) /lib64/ld-linux-x86-64.so.2 (0x003e17b0) *** /var/MailRoot/bin/XMail127v3 -Md -Ms /var/MailRoot -Sl -Pl -Cl \ -SI 0.0.0.0:8025 -SI 0.0.0.0:25 \ -PI 0.0.0.0:8110 -PI 0.0.0.0:110 \ -BI 0.0.0.0:995 \ -XI 0.0.0.0:465 \ -Ql -Ln 256 -Pt 60 -St 150 -F- -SX 2048 -Mx 64 \ -Qn 64 -PX 2048 -Qr 1 -Qt 1 -QT 150 -Qg -CX 256 *** Original Message From: "U.Mutlu" To: t...@os.st, "XMail Users Mailing List" Sent: Tue, May 26, 2015, 1:06 PM Subject: Re: [xmail] BUG : XMAIL 1.27 with SSL Oversea Site via xmail wrote on 05/26/2015 05:14 AM: No Log Error. No Fatal Error log. No Core Dump It's not RAM or HW problem. It's XMAIL internal issue, because when I switch to stunnel, all problems solved. And then when I test K9-MAIL with XMAIL+STUNNEL, K9 Client will not have any JAVA somethings Exception Error Two tests 1. Very Important Thing in Xmail 1.27 Native SSL If the emailbox has many emails such as 1, Microsoft Outlook with SSL or NON-SSL download them, it WILL NOT have this issue. Some Tencent Foxmail for Windows, after finish ssl session, it will take the server be FIN_WAIT2 (some version, not all) But K9-MAIL for Android, Download first 1000, it will appear the error in K9-MAIL randomly. 2. Xmail 1.27 + Stunnel Proxy If the emailbox has many
Re: [xmail] BUG : XMAIL 1.27 with SSL
Oversea Site via xmail wrote on 05/26/2015 07:56 AM: Anyway I will test this issue in other linux distribution in VM, such as CentOS. (You can give me any linux distribution, let's build the same platform in VM. I use virtualbox. If you could test it with Debian 7 (wheezy) i386, that would make it easier for me to track the bug. -- Thx Uenal ___ xmail mailing list xmail@xmailserver.org http://xmailserver.org/mailman/listinfo/xmail
Re: [xmail] BUG : XMAIL 1.27 with SSL
U.Mutlu via xmail wrote on 05/26/2015 07:06 AM: Oversea Site via xmail wrote on 05/26/2015 05:14 AM: No Log Error. No Fatal Error log. No Core Dump It's not RAM or HW problem. It's XMAIL internal issue, because when I switch to stunnel, all problems solved. And then when I test K9-MAIL with XMAIL+STUNNEL, K9 Client will not have any JAVA somethings Exception Error Two tests 1. Very Important Thing in Xmail 1.27 Native SSL If the emailbox has many emails such as 1, Microsoft Outlook with SSL or NON-SSL download them, it WILL NOT have this issue. Some Tencent Foxmail for Windows, after finish ssl session, it will take the server be FIN_WAIT2 (some version, not all) But K9-MAIL for Android, Download first 1000, it will appear the error in K9-MAIL randomly. 2. Xmail 1.27 + Stunnel Proxy If the emailbox has many emails such as 1, Microsoft Outlook with SSL or NON-SSL download them, it WILL NOT have this issue. Some Tencent Foxmail for Windows, after finish ssl session, it WILL NOT have any problem. Server WILL NOT TAKE FIN_WAIT_2 K9-MAIL for Android, Download first 1000, it will NOT DISPLAY any error when it finished download emails. I would need full description about the server HW and SW (OS version, xmail start params etc., any xmail patches applied etc.). There is of course also the possibility that the bug lies in the said mobile client program, especially if it goes online via unstable mobile links as it seems to be a link latency problem, or the client does not close the socket cleanly. Take a look at these: https://github.com/Automattic/socket.io/issues/1380 http://stackoverflow.com/questions/5328155/preventing-fin-wait2-when-closing-socket Check this too: # cat /proc/sys/net/ipv4/tcp_fin_timeout 60 The following page is informative too: http://httpd.apache.org/docs/2.0/misc/fin_wait_2.html -- cu Uenal ___ xmail mailing list xmail@xmailserver.org http://xmailserver.org/mailman/listinfo/xmail
Re: [xmail] BUG : XMAIL 1.27 with SSL
Oversea Site via xmail wrote on 05/26/2015 05:14 AM: No Log Error. No Fatal Error log. No Core Dump It's not RAM or HW problem. It's XMAIL internal issue, because when I switch to stunnel, all problems solved. And then when I test K9-MAIL with XMAIL+STUNNEL, K9 Client will not have any JAVA somethings Exception Error Two tests 1. Very Important Thing in Xmail 1.27 Native SSL If the emailbox has many emails such as 1, Microsoft Outlook with SSL or NON-SSL download them, it WILL NOT have this issue. Some Tencent Foxmail for Windows, after finish ssl session, it will take the server be FIN_WAIT2 (some version, not all) But K9-MAIL for Android, Download first 1000, it will appear the error in K9-MAIL randomly. 2. Xmail 1.27 + Stunnel Proxy If the emailbox has many emails such as 1, Microsoft Outlook with SSL or NON-SSL download them, it WILL NOT have this issue. Some Tencent Foxmail for Windows, after finish ssl session, it WILL NOT have any problem. Server WILL NOT TAKE FIN_WAIT_2 K9-MAIL for Android, Download first 1000, it will NOT DISPLAY any error when it finished download emails. I would need full description about the server HW and SW (OS version, xmail start params etc., any xmail patches applied etc.). There is of course also the possibility that the bug lies in the said mobile client program, especially if it goes online via unstable mobile links as it seems to be a link latency problem, or the client does not close the socket cleanly. Take a look at these: https://github.com/Automattic/socket.io/issues/1380 http://stackoverflow.com/questions/5328155/preventing-fin-wait2-when-closing-socket cu Uenal Original Message From: "U.Mutlu" To: t...@os.st, "XMail Users Mailing List" Sent: Tue, May 26, 2015, 10:11 AM Subject: Re: [xmail] BUG : XMAIL 1.27 with SSL Oversea Site via xmail wrote on 05/25/2015 09:23 PM: I found the bug on XMAIL 1.27 with SSL (POP3S is main), SSMTP service maybe has a same bug too 1. it will have too many FIN_WAIT_2 on POP3S with two many requests, It is because of some non-popular WINDOWS CLIENT e.g. tencent foxmail , outlook will not have this issue. 2. If you're using XMAIL'S internal SSL POP3S with any openssl lib. (tested 0.9.8za 1.0.0, tested gcc 4.0.2/4.0.4/4.3.6) and the email account contain too many emails (1000-2000 emails), the mobile client POP3 such as Samsung internal Email Client is using SSL, K9-MAIL on any android devices using SSL, some iPhone devices (connect with XMAIL SSL port 995), will not download them successfully (Android Error - Java Exception Error) and some iPhone device will stuck with it, it will not download all emails successfully. But if you disable XMAIL'S internal POP3S with -B- startup option, and switch stunnel 5.17 port 995 tunnel to port 110, it will solve this problem and kick out FIN_WAIT_2 * stunnel 5.17 config * [pop3s] accept = 995 connect = 110 cert = /etc/stunnel/stunnel.pem *** What do the logs (server, client) say about the errors? I think you need a faster server HW and/or need to optimize the resources, for example the number of threads, and possibly also RAM. An excerpt from the doc: " If you have an heavily loaded server remember to setup the best number of XMAIL threads by specifying the '-Qn nthreads' option (you must do some tentatives to find the best value for your needs). Also you can limit the number of SMTP, POP3 and CTRL service threads by specifying the options '-SX maxthreads', '-PX maxthreads' and '-CX maxthreads'. " If there are mail filters, you can put them into a ramdisk and set PATH accordingly, or use an absolute path to them. See also XMAIL_TEMP in the doc. In task manager / system monitor you can see how the resource usage is, especially CPU, memory, and network. -- cu Uenal ___ xmail mailing list xmail@xmailserver.org http://xmailserver.org/mailman/listinfo/xmail ___ xmail mailing list xmail@xmailserver.org http://xmailserver.org/mailman/listinfo/xmail
Re: [xmail] BUG : XMAIL 1.27 with SSL
Oversea Site via xmail wrote on 05/25/2015 09:23 PM: I found the bug on XMAIL 1.27 with SSL (POP3S is main), SSMTP service maybe has a same bug too 1. it will have too many FIN_WAIT_2 on POP3S with two many requests, It is because of some non-popular WINDOWS CLIENT e.g. tencent foxmail , outlook will not have this issue. 2. If you're using XMAIL'S internal SSL POP3S with any openssl lib. (tested 0.9.8za 1.0.0, tested gcc 4.0.2/4.0.4/4.3.6) and the email account contain too many emails (1000-2000 emails), the mobile client POP3 such as Samsung internal Email Client is using SSL, K9-MAIL on any android devices using SSL, some iPhone devices (connect with XMAIL SSL port 995), will not download them successfully (Android Error - Java Exception Error) and some iPhone device will stuck with it, it will not download all emails successfully. But if you disable XMAIL'S internal POP3S with -B- startup option, and switch stunnel 5.17 port 995 tunnel to port 110, it will solve this problem and kick out FIN_WAIT_2 * stunnel 5.17 config * [pop3s] accept = 995 connect = 110 cert = /etc/stunnel/stunnel.pem *** What do the logs (server, client) say about the errors? I think you need a faster server HW and/or need to optimize the resources, for example the number of threads, and possibly also RAM. An excerpt from the doc: " If you have an heavily loaded server remember to setup the best number of XMAIL threads by specifying the '-Qn nthreads' option (you must do some tentatives to find the best value for your needs). Also you can limit the number of SMTP, POP3 and CTRL service threads by specifying the options '-SX maxthreads', '-PX maxthreads' and '-CX maxthreads'. " If there are mail filters, you can put them into a ramdisk and set PATH accordingly, or use an absolute path to them. See also XMAIL_TEMP in the doc. In task manager / system monitor you can see how the resource usage is, especially CPU, memory, and network. -- cu Uenal ___ xmail mailing list xmail@xmailserver.org http://xmailserver.org/mailman/listinfo/xmail
Re: [xmail] Tip: Speeding up xmail by pointing envvar XMAIL_TEMP to a ramdisk
Bart Mortelmans via xmail wrote on 05/25/2015 02:24 PM: Are you sure this will actually speed up XMail? Have you been able to compare, specifically for XMail? It will help for files xmail creates in /tmp and doesn't move to an other drive (for example when the realtime filters get fired during mail acceptance, the passed file always lies in /tmp). Since a /tmp on a ramdisk is ultra-fast I concluded that this will speed things up. But I'm not sure yet if xmail moves that file to main disk. As far as I understand how XMail processes e-mails, it actually only creates a file one time and then moves that file around between temp-folder, spool folder and local POP3 folders. Moving a file on the same filesystem means the file isn't re-rewritten. If you move a file between different filesystems, it does need to be re-written. That could actually mean that having the temp folder on a RAM-disk might slow things down if there would be an event in which XMail moves a file from spool to temp to local POP3 folder. When they are all on the same filesystem, this would mean the file would need to be written only once (and then moved around). If the temp folder is on a different file-system, this would cause the file to be writting to the hard drive twice and to the RAM-drive once. Hmm. yes, true. But there are use cases where ramdisk will be w/o question useful, for example starting the filter-programs from ramdisk. All mails will not only pass through the temp-folder, but will also end up in the spool-folder. If most mails only pass through the server (and don't need to be stored locally) then you might be able to see a speed improvement by also placing the spool folder on a ramdisk. But this would mean that all mails pending delivery are lost at server restart (so this isn't really recommended). Yes, true, putting spool on ramdisk is not that a good idea :-) cu Uenal Sincerely, Bart Mortelmans U.Mutlu via xmail schreef op 25/05/15 om 01:09: Tip: Speeding up xmail by pointing envvar XMAIL_TEMP to a ramdisk (here on Linux using a 64 MB ramdisk mounted to /mnt/ramdisk ) mkdir -p /mnt/ramdisk in /etc/fstab: tmpfs /mnt/ramdisk tmpfs nodev,size=64M 0 0 then either reboot or use this cmd: mount /mnt/ramdisk in xmail start script: export XMAIL_ROOT=/var/MailRoot export XMAIL_TEMP=/mnt/ramdisk export PATH=$XMAIL_ROOT/bin:/usr/local/sbin:/usr/local/bin:/sbin:/bin:/usr/sbin:/usr/bin ... and restart xmail. The ramdisk size should be bigger than "MaxMessageSize" (unit KB) in server.tab. This will speed up mail processing (at least incoming mails) significantly. Here's more info about speed gains: http://www.jamescoyle.net/knowledge/951-the-difference-between-a-tmpfs-and-ramfs-ram-disk "The major benefit to memory based file systems is that they are very fast – 10s of times faster than modern SSDs. Read and write performance is massively increased for all workload types." One can even copy the mostly called programs (for example filters) onto the ramdisk, but then one must use such a PATH line (instead of the above one): export PATH=/mnt/ramdisk:$XMAIL_ROOT/bin:/usr/local/sbin:/usr/local/bin:/sbin:/bin:/usr/sbin:/usr/bin ___ xmail mailing list xmail@xmailserver.org http://xmailserver.org/mailman/listinfo/xmail
[xmail] Tip: Speeding up xmail by pointing envvar XMAIL_TEMP to a ramdisk
Tip: Speeding up xmail by pointing envvar XMAIL_TEMP to a ramdisk (here on Linux using a 64 MB ramdisk mounted to /mnt/ramdisk ) mkdir -p /mnt/ramdisk in /etc/fstab: tmpfs /mnt/ramdisk tmpfs nodev,size=64M 0 0 then either reboot or use this cmd: mount /mnt/ramdisk in xmail start script: export XMAIL_ROOT=/var/MailRoot export XMAIL_TEMP=/mnt/ramdisk export PATH=$XMAIL_ROOT/bin:/usr/local/sbin:/usr/local/bin:/sbin:/bin:/usr/sbin:/usr/bin ... and restart xmail. The ramdisk size should be bigger than "MaxMessageSize" (unit KB) in server.tab. This will speed up mail processing (at least incoming mails) significantly. Here's more info about speed gains: http://www.jamescoyle.net/knowledge/951-the-difference-between-a-tmpfs-and-ramfs-ram-disk "The major benefit to memory based file systems is that they are very fast – 10s of times faster than modern SSDs. Read and write performance is massively increased for all workload types." One can even copy the mostly called programs (for example filters) onto the ramdisk, but then one must use such a PATH line (instead of the above one): export PATH=/mnt/ramdisk:$XMAIL_ROOT/bin:/usr/local/sbin:/usr/local/bin:/sbin:/bin:/usr/sbin:/usr/bin -- cu Uenal ___ xmail mailing list xmail@xmailserver.org http://xmailserver.org/mailman/listinfo/xmail
[xmail] HOWTO quick-testing spamassassin (spamd + spamc) at home
Hi Spros & All, this maybe might interest some of you: HOWTO quick-testing spamassassin (spamd + spamc) at home spamd is the daemon program (runs in background) spamc is the client pgm to interact with spamd 1.) install spamassassin pkg (Debian-Linux: apt-get install spamassassin) 2.) enable the spamd daemon (in file /etc/default/spamassassin set ENABLED=1) 3.) start spamd ("/etc/init.d/spamassassin start" or "service spamassassin start" depending on your init-system (syvinit/upstart/systemd)) 4.) in your mailclient save a single mail to disk for testing 5.) spamc -R < mailfile.eml This gives the result of spam testing Example: " $ spamc -R < test3.eml -0.5/5.0 Spam detection software, running on the system "xx", has NOT identified this incoming email as spam. ... Content analysis details: (-0.5 points, 5.0 required) pts rule name description -- -- 0.0 URIBL_BLOCKED ADMINISTRATOR NOTICE: The query to URIBL was blocked. See http://wiki.apache.org/spamassassin/DnsBlocklists#dnsbl-block for more information. [URIs: srpsecure.de] 0.0 TVD_RCVD_SPACE_BRACKET No description available. -0.6 RP_MATCHES_RCVDEnvelope sender domain matches handover relay domain 0.0 T_FRT_CONTACT BODY: ReplaceTags: Contact 0.0 HTML_MESSAGE BODY: HTML included in message 0.0 DC_PNG_UNO_LARGO Message contains a single large inline gif " After testing you can stop spamd, deaktive it or uninstall if desired. In real life one would install spamassassin on the mailserver and create a filter to pass the to be checked mail to it, and analyse its return value etc... More info @ http://wiki.apache.org/spamassassin/ cu Uenal Spyros Tsiolis via xmail wrote on 05/24/2015 06:34 PM: I was introduced to XMail by a colleague of mine. At first it seemed cryptic and it still is for some stuff I am interested in. I've been using it since 2003 and never looked back. --------------- On Sun, 24/5/15, U.Mutlu via xmail wrote: Subject: Re: [xmail] new maintainer for the 1.27 branch To: "Bart Mortelmans" , "XMail Users Mailing List" Date: Sunday, 24 May, 2015, 19:07 Hi Bart & All, Bart Mortelmans via xmail wrote on 05/16/2015 10:27 PM: > I have been using XMail for what must have been almost 15 years. I still > prefer XMail for anything that does not need IMAP. I have no experience yet with Dovecot, but what does it mean when they say "If you already have XMail Setup and just want to add in IMAP support See : dovecot-setup.html" on this webpage: http://xmail.nomadcf.com/xmail-dovecot I think what it's trying to say is : "if you have already XMail setup for pop3/smtp, all you need is dovecot for lmtp/imap. I have working installations with XMail / dovecot / Horde and it works like a dream. No problems there. That has been for about seven years; Since 2008 I think. > I made some small changes to the code. If anybody would be willing to pick > up XMail and start development again, I hope that these changes can also > get into the new version. If you need more details, let me know! Thx, since Davide is not reachable for a very long time now (3+ yrs?), WOW ! 3 Years ? I know he has a life (no pun intended) apart from his XMail project. I also know he has developed other stuff (see here : http://xmailserver.org/davide.html) but I never expected for him to just abandon XMail. It's a good project. Why would he do this ? I hereby accounce my general interest in bugfixing and adding new features (patches and wishes of users) into the current 1.27 branch, if time permits. I'll also setup an svn repository. There are many nice things one could add to XMail. However I am not a developer and I can only propose stuff. One feature I longed for is the ability to view e-mail traffic in real-time and also XMail be able to log incoming and outgoing messages. The only problem is: at this time I can't support any other OS beyond Linux. will brand the new version something like "v1.27.n.linux" where "n" is a counter. > - Change the default error message for a failed pre-date or post-data > filter to code 451 (by default this would otherwise be 554 which means > there won’t be an other attempt). For me 451 makes much more sense. Can you explain how this is meant? Is it the return code of the filter program like discussed here? : http://xmailforum.marketmix.com/index.php?showtopic=3362 > - Added basic support for LMTP. That way I can have XMail deliver mails to > dovecot directly. For this, I just had to add the LHLO m
Re: [xmail] new maintainer for the 1.27 branch
Hi Bart & All, Bart Mortelmans via xmail wrote on 05/16/2015 10:27 PM: I have been using XMail for what must have been almost 15 years. I still prefer XMail for anything that does not need IMAP. I have no experience yet with Dovecot, but what does it mean when they say "If you already have XMail Setup and just want to add in IMAP support See : dovecot-setup.html" on this webpage: http://xmail.nomadcf.com/xmail-dovecot I made some small changes to the code. If anybody would be willing to pick up XMail and start development again, I hope that these changes can also get into the new version. If you need more details, let me know! Thx, since Davide is not reachable for a very long time now (3+ yrs?), I hereby accounce my general interest in bugfixing and adding new features (patches and wishes of users) into the current 1.27 branch, if time permits. I'll also setup an svn repository. The only problem is: at this time I can't support any other OS beyond Linux. I will brand the new version something like "v1.27.n.linux" where "n" is a counter. - Change the default error message for a failed pre-date or post-data filter to code 451 (by default this would otherwise be 554 which means there won’t be an other attempt). For me 451 makes much more sense. Can you explain how this is meant? Is it the return code of the filter program like discussed here? : http://xmailforum.marketmix.com/index.php?showtopic=3362 - Added basic support for LMTP. That way I can have XMail deliver mails to dovecot directly. For this, I just had to add the LHLO message next to the HELO and EHLO. This works for me, as with my configuration there shouldn’t be any event in which XMail tries to deliver one e-mail to multiple recipients via LTMP. I’m not sure if there might be other configurations for which more work would be needed for LMTP to work. This sounds not trivial :-) I must admit I have no experience yet with LMTP, but I saw wikipedia has a page about it: https://en.wikipedia.org/wiki/Local_Mail_Transfer_Protocol https://tools.ietf.org/html/rfc2033 - Made XMail log a line to SMAIL log in case of delivery failure (posted details on this in an e-mail to this mailing list very recently). Yeah, I would need that feature too. I recently added this feature into xmail: spamassassin (spamd via spamc) reports, with default config, to every mail received by xmail the following bad score of 1.3: "1.3 RDNS_NONE Delivered to internal network by a host with no rDNS" After much research and experimenting I figured out what the reason is: it is the "Received:" line xmail prepends to the mail depending on "ReceivedHdrType" in server.tab. There are choices from 0 to 4, and I added one more (5), and now spamassassin no more gives the above spam-score. If anybody is taking feature requests: I would be interested in a way of XMail talking to a “filter” via a socket. Now, for every e-mail to be filtered, a process has to be started. It would be much more efficient if XMail could talk a filter-server via a socket. That filter server could be anything, but maybe we could even get it to talk directly to spamd and clamd… Yes, that's a good idea. I'll study the xmail sources. -- Thx Uenal ___ xmail mailing list xmail@xmailserver.org http://xmailserver.org/mailman/listinfo/xmail
[xmail] thx to xmail and its author
Hi folks, I'm using xmail for a long time now (maybe 7 yrs), and I'm still very satisfied with it. OTOH I must admit I'm using it only for a small number of users, and I've used (before xmail) only one other mail-system (it was qmail iirc). I'm using a self-compiled version of xmail (the crippled version in the debian repository I tried about 2 yrs ago was unusable due to cfg-files spread around to many OS-system dirs; I like it compact under a single app-dir and its subdirs). The only thing I don't like much is the usage of TAB as delimiter in the cfg-files. What I would like to ask is: - What are the shortcomings or missing features other users experience with xmail? - What about the new developments regarding ssl and crypto since Snowden, and their relevance to xmail? - What new features should xmail have implemented? -- Thx Uenal ___ xmail mailing list xmail@xmailserver.org http://xmailserver.org/mailman/listinfo/xmail
[xmail] non-delivered mails not listed in the logs
Hi, anybody left here? :-) Can anybody confirm this? : It seems xmail does not make a log entry for a non-deliverable mail. Though it sends an error mail to the sender (Subject "Error sending message ..."). This is not optimal, since the mail admin should have the possibility to inform himself about such problems from the logs. Thx ___ xmail mailing list xmail@xmailserver.org http://xmailserver.org/mailman/listinfo/xmail