Toutes les versions
1.0.0-beta.5
This release brings 1 improvement and 12 fixes.
Améliorations
- Migration scans of HestiaCP, Plesk and other panels read the panel's own figures. A migration scan now takes what the source panel already records instead of measuring every account: on HestiaCP each user's disk usage and database count from the user list it reads once; on Plesk every subscription with its system user, status, plan, PHP version, disk usage and database count from one query on Plesk's own database, so a server with hundreds of subscriptions no longer runs several Plesk commands for each; on DirectAdmin, CWP and CyberPanel every account's database count in one query, and on CyberPanel every site's mailbox count from the panel's mail records in one query. Plesk accounts now keep their PHP version, the database and mailbox counts of CyberPanel sites and the database counts of CloudPanel sites are right again, CyberPanel's packages are listed for plan mapping, and CWP accounts keep the PHP version CWP records for them. Accounts the panel has no disk figure for are still measured as before.
Corrections de bugs
- Migrations from DirectAdmin, Plesk and CyberPanel bring each site's own databases. Databases on DirectAdmin, Plesk and CyberPanel servers are now read with the login each panel keeps for its database server, so they are copied instead of failing or being left out. Plesk and CyberPanel do not start database names with the account name: the panel's own record of the subscription or website now decides which databases and users come over, each gets the account's name as its prefix here with the same password, and the site's wp-config.php is pointed at the new names; a Plesk database kept on another database server is named in the migration warnings. A CyberPanel migration copies each account's own website instead of the server's first one. A database server that refuses the panel's login, a database name this server cannot take, and CloudPanel (which keeps no readable record of a site's databases) are now reported in the account's migration warnings instead of passing silently.
- Migrations from DirectAdmin and HestiaCP keep each site's PHP version. A site migrated from DirectAdmin now gets the PHP version of the CustomBuild slot its domain uses (a domain without a choice runs the first slot), instead of the server's default. A site migrated from HestiaCP gets the version of its domain's PHP-FPM backend template (PHP-8_2 gives 8.2); a domain on the default template keeps this server's default. The version is used when it runs on this server, as before.
- WordPress permalinks and .htaccess rewrites work on OpenLiteSpeed and LiteSpeed sites. Sites served by OpenLiteSpeed now apply the rewrite rules of their .htaccess files: WordPress pages, posts and category archives under pretty permalinks open instead of answering 404, and other .htaccess rewrite rules take effect as they do on Apache. On LiteSpeed Enterprise the same .htaccess directives Apache sites may use are read (before, only access rules were). Existing sites are rewritten once at the first start after the update; the redirects set in the panel stay as they are.
- Sites moved to OpenLiteSpeed or LiteSpeed keep their PHP extensions, MySQL included. Switching the server to OpenLiteSpeed or LiteSpeed Enterprise now installs LiteSpeed's PHP for every PHP version with the same extensions the PHP-FPM version has (MySQL, mbstring, XML, GD, intl, OPcache and the rest). Before, only the bare PHP build was installed, so a WordPress or other MySQL site on PHP 8.4 or 8.2 answered with a 500 error after the switch. A server already on OpenLiteSpeed or LiteSpeed gets the missing extensions in the background shortly after the first start following the update, then the web server restarts gracefully.
- Switching to OpenLiteSpeed no longer ends with every site on its default page. When two changes reached the LiteSpeed web server a few seconds apart (switching it on, a WAF change, an account's PHP version, restarts from the Services page), the second could stop the server while it was still starting: the switch still said it had succeeded, but every site showed OpenLiteSpeed's welcome page, and account changes answered with an error. Each action now waits until the server is fully up, and a switch to OpenLiteSpeed or LiteSpeed Enterprise is only reported as done once the new server is the one answering for the sites; otherwise the sites go back to their previous web server and the switch reports the failure.
- A LiteSpeed Enterprise activation that hangs says why and leaves OpenLiteSpeed usable. When LiteSpeed's own installer cannot reach one of LiteSpeed's download hosts during a LiteSpeed Enterprise activation, WHost now stops it after 5 minutes together with every download it started, and the activation error names what the installer was waiting on and its last line of output (before, a download kept running in the background and the error only said the command timed out). After such a failed attempt, switching OpenLiteSpeed on works again: OpenLiteSpeed's package, WAF module and service are put back over the files the installer left.
- A removed OpenLiteSpeed or LiteSpeed site is gone from WHost's configuration backup too. WHost keeps a backup copy of the OpenLiteSpeed / LiteSpeed server configuration (httpd_config.conf.bak, httpd_config.xml.bak). Before, the write that removed a site (an account deleted on OpenLiteSpeed, or every site when OpenLiteSpeed or LiteSpeed is switched off) left the copy as it was before the removal, so it went on naming the removed site; now the copy keeps the result of a removal and never names a site the server no longer hosts. Other writes keep the previous content as before. LiteSpeed's own history files (httpd_config.conf0 and its revisions, httpd_config.conf.txt) are LiteSpeed's and are left as they are.
- Redirects set in the panel work on OpenLiteSpeed and LiteSpeed sites. A redirect set in the panel (Domains -> Redirects) for a site on OpenLiteSpeed or LiteSpeed Enterprise answered 404 instead of redirecting. Redirects are now written where the server applies them to the whole path, ahead of the site's folders and its .htaccess rules, as on Nginx; sites already on OpenLiteSpeed or LiteSpeed get it at the first agent start after the update (one render, one configuration test, one graceful restart).
- Switching from OpenLiteSpeed to LiteSpeed Enterprise works. The LiteSpeed Enterprise card offers a switch while OpenLiteSpeed serves the sites. That switch never finished: LiteSpeed's installer waited five minutes for the ports OpenLiteSpeed holds and the activation failed. Now OpenLiteSpeed is switched off first, the sites go back to Nginx + Apache and stay up there while LiteSpeed Enterprise is installed, then move into its configuration; if the installation fails they stay on Nginx + Apache (switching OpenLiteSpeed on again takes them back).
- Reloading LiteSpeed on the Services page no longer holds the panel's other requests. On a server running OpenLiteSpeed or LiteSpeed Enterprise, a restart or reload from the Services page waits until the new server generation is ready (a few seconds, longer with the WAF rule set). That wait held every other request to the agent, and a request made meanwhile could fail with a gateway timeout (504). The wait now runs aside and the panel keeps answering.
- The Ubuntu installation stops with the real reason when a package source cannot be added. On Ubuntu the installer adds the PHP (ondrej/php) and Python (deadsnakes) package sources from Launchpad. When Launchpad did not answer, the installer waited about two minutes, went on without the PHP source and then stopped at "Unable to locate package", which read like a missing package. Now each attempt is limited to 150 seconds and made up to three times; if the source still cannot be added, the installer stops at that step and says so ("Could not add ppa:ondrej/php from Launchpad ... (last try: ...)"), with the command's own output in the install log. It also checks that the new source is read and offers the PHP version before it goes on.
- Progress bars tell screen readers their percentage. The panel's progress bars (disk and resource usage, OPcache memory, Server Information, update and upload progress) now carry their value for assistive technology: a screen reader announces the percentage instead of an empty "progress bar". A usage above its limit still shows a full bar and is announced as 100%. The bars look the same as before.
Dans cette version
- Publié le
- Redémarrage
- Requis après la mise à jour