{"id":164053,"date":"2026-03-13T11:11:58","date_gmt":"2026-03-13T09:11:58","guid":{"rendered":"https:\/\/prohoster.info\/blog\/novosti-interneta\/predlozhenie-po-perevodu-sistemnyh-logov-lastlog-btmp-utmp-i-wtmp-na-ispolzovanie-sqlite"},"modified":"2026-03-13T11:11:58","modified_gmt":"2026-03-13T09:11:58","slug":"predlozhenie-po-perevodu-sistemnyh-logov-lastlog-btmp-utmp-i-wtmp-na-ispolzovanie-sqlite","status":"publish","type":"post","link":"https:\/\/prohoster.info\/hu\/blog\/news\/predlozhenie-po-perevodu-sistemnyh-logov-lastlog-btmp-utmp-i-wtmp-na-ispolzovanie-sqlite","title":{"rendered":"Javaslat a rendszernapl\u00f3k, a lastlog, btmp, utmp \u00e9s wtmp SQLite-ra val\u00f3 \u00e1tt\u00e9r\u00e9s\u00e9re.","gt_translate_keys":[{"key":"rendered","format":"text"}]},"content":{"rendered":"
A linux-api mailing list proposal (RFC) has been put forward for discussion to replace the outdated binary formats of the system logs lastlog, btmp, utmp, and wtmp with new shared libraries that use SQLite as a backend. This initiative aims to address accumulated issues, including the overflow of 32-bit time counters in 2038, lack of extensibility, low query performance, and the absence of atomicity during writes. <\/p>\n
Currently, the following binary files with a fixed structure are used in Linux to store session data and authentication attempts: <\/p>\n
The data format of these files was developed several decades ago and has a number of fundamental limitations: <\/p>\n
\n
A «tv_sec» mez\u0151 az «utmpx» strukt\u00far\u00e1ban \u00e9s az «ll_time» mez\u0151 a «lastlog» eset\u00e9ben «int32_t» t\u00edpussal b\u00edr, \u00e9s az id\u0151sz\u00e1m\u00edt\u00e1si sz\u00e1ml\u00e1l\u00f3k \u00e9rt\u00e9ke 2038. janu\u00e1r 19-\u00e9n fog t\u00fall\u00e9pni. Az ABI‑kompatibilit\u00e1s k\u00f6vetelm\u00e9nyei miatt ezek a mez\u0151k m\u00e9g 64 bites rendszereken is 32 bitesek maradnak, \u00edgy a probl\u00e9ma minden Linux telep\u00edt\u00e9st \u00e9rinteni fog.\n
The fixed size of records does not allow for the addition of new fields (such as container ID, service name, IP address) without completely replacing the format and recompiling all utilities.\n
The last, lastb, who, and lastlog utilities are forced to linearly scan the contents of the files. With large log sizes, the lack of indexes for effectively filtering records leads to an unacceptable load on input\/output systems and delays in query execution.\n
Writing to a binary file is not an atomic operation. In the event of a failure, the write may become partially corrupted.\n
To avoid conflicts during simultaneous writing to the logs by multiple processes (such as sshd and login), flock locks are used, which do not guarantee atomicity and can lead to deadlocks. <\/ul>\n
Az RFC szerz\u0151je javasolja a bin\u00e1ris form\u00e1tumok teljes elhagy\u00e1s\u00e1t a SQLite-ot haszn\u00e1l\u00f3, speci\u00e1lis megosztott k\u00f6nyvt\u00e1rak jav\u00e1ra. Minden napl\u00f3z\u00e1si t\u00edpushoz k\u00fcl\u00f6n k\u00f6nyvt\u00e1r k\u00e9sz\u00fcl egys\u00e9ges C interf\u00e9sszel: liblastlog2, libbtmp2, libutmp2 \u00e9s libwtmp2. Minden k\u00f6nyvt\u00e1r adatb\u00e1zisokkal dolgozik, amelyek s\u00e9m\u00e1i 64-bites id\u0151b\u00e9lyegeket (INTEGER t\u00edpus) \u00e9s indexeket tartalmaznak felhaszn\u00e1l\u00f3 \u00e9s id\u0151 szerint. Lehet\u0151s\u00e9g van \u00faj mez\u0151k hozz\u00e1ad\u00e1s\u00e1ra an\u00e9lk\u00fcl, hogy megszak\u00edtan\u00e1k a kompatibilit\u00e1st (ALTER TABLE seg\u00edts\u00e9g\u00e9vel). <\/p>\n
Az SQLite haszn\u00e1lat\u00e1nak \u00e9rvek k\u00f6z\u00f6tt szerepel a 64-bites INTEGER t\u00edpus alkalmaz\u00e1sa az epoch\u00e1lis id\u0151 t\u00e1rol\u00e1s\u00e1ra, indexek felhaszn\u00e1l\u00e1sa az I\/O cs\u00f6kkent\u00e9se \u00e9rdek\u00e9ben, a rekordok szelekt\u00edv el\u00e9r\u00e9s\u00e9nek lehet\u0151s\u00e9ge a teljes beolvas\u00e1s helyett, \u00faj mez\u0151k hozz\u00e1ad\u00e1s\u00e1nak lehet\u0151s\u00e9ge a megl\u00e9v\u0151 bejegyz\u00e9sek m\u00f3dos\u00edt\u00e1sa n\u00e9lk\u00fcl, az ACID tranzakci\u00f3k t\u00e1mogat\u00e1sa, a WAL (Write-Ahead Logging) \u00fczemm\u00f3d a z\u00e1rol\u00e1s n\u00e9lk\u00fcli egyidej\u0171 hozz\u00e1f\u00e9r\u00e9shez, valamint az SQLite megb\u00edzhat\u00f3s\u00e1g\u00e1nak megfelel\u0151en tesztelt m\u0171k\u00f6d\u00e9se. <\/p>\n
A z\u00f6kken\u0151mentes \u00e1tmenet biztos\u00edt\u00e1sa \u00e9rdek\u00e9ben a «dupla \u00edr\u00e1s» (dual-write) strat\u00e9gi\u00e1t javasoljuk: <\/p>\n
\n
Azok a programok, amelyek bin\u00e1ris f\u00e1jlokba \u00edrnak (login, sshd, sudo, cron stb.), m\u00f3dos\u00edt\u00e1sra ker\u00fclnek, hogy egyszerre \u00edrjanak a r\u00e9gi bin\u00e1ris f\u00e1jlba \u00e9s az \u00faj SQLite-adatb\u00e1zisba a megfelel\u0151 k\u00f6nyvt\u00e1ron kereszt\u00fcl.\n
\u00daj verzi\u00f3k k\u00e9sz\u00fclnek az olyan seg\u00e9dprogramokb\u00f3l (last2, lastb2, who2, lastlog2), amelyek az SQLite-adatb\u00e1zisokb\u00f3l olvasnak, indexek haszn\u00e1lat\u00e1val a gyors m\u0171k\u00f6d\u00e9s \u00e9rdek\u00e9ben. A r\u00e9gi seg\u00e9dprogramok tov\u00e1bbra is m\u0171k\u00f6dnek a kor\u00e1bbi f\u00e1jlokkal.\n
N\u00e9h\u00e1ny \u00e9v m\u00falva, amikor a t\u00f6bbs\u00e9g rendszere friss\u00fcl, a r\u00e9gi form\u00e1tumokba val\u00f3 \u00edr\u00e1s t\u00e1mogat\u00e1sa ki lehet kapcsolva, \u00e9s a r\u00e9gi seg\u00e9dprogramok — elavultnak nyilv\u00e1n\u00edthat\u00f3k. <\/ul>\n
K\u00e9rd\u00e9sek, amelyek tov\u00e1bbi megvitat\u00e1sra v\u00e1rnak: <\/p>\n
\n
R\u00e9szleges k\u00f6nyvt\u00e1rak sz\u00e9tv\u00e1laszt\u00e1s\u00e1nak vagy egyes\u00edt\u00e9s\u00e9nek c\u00e9lszer\u0171s\u00e9ge (p\u00e9ld\u00e1ul libsession2).\n