severity 610185 wishlist thanks [Mike Bird] wrote: > Blindly enabling insserv can result in unbootable systems or systems > which boot but don't start all services. sysv-rc.postinst currently > states that enabling insserv is "recommended", with no mention of > the possible severe consequences.
The current template ask users /usr/share/doc/sysv-rc/README.Debian for more information, which is the place I believe it is best to add more information on the consequences. We used to have more text in the template, but was adviced by the translators to trim down the text and move it to the README. Note that the legacy boot ordering is filled with hard to fix bugs (incorrect ordering), and this is the reason for recommending to move to the new dependency based boot sequencing. The improved speed is a by-product and not the main advantage of dependency based boot sequencing. > This bug is critical rather than grave because the lost data - the > Snn and Knn values - cannot be recovered without restoring a backup. The information is not really lost. The values present during conversion is saved in /var/lib/update-rc.d/, and restored when migrating to file-rc. > (In fact it's slightly worse than that - if you merely restore /etc > insserv will eat it again. You have to use undocumented magic to > prevent this.) I assume you mean the /etc/init.d/.legacy-bootordering when you talk about undocumented magic. I thought it was documented, but it might be an idea to verify this. Given your strong feelings about boot script ordering, I suspect you are better off using file-rc until it is adjusted to use dependency based boot sequencing. I am convinced the static sequence numbers are slowly going to become unmaintained in Debian, as more and more stop caring about their values, so it might be a good idea to adjust to the dynamic dependency based boot sequencing sooner than later. I believe this issue at most is a wishlist issue, and I note at least one of the release managers believe this is a none-issue (he closed the bug), so I adjust the severity. Happy hacking, -- Petter Reinholdtsen -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
