Re: Sun OS 4.03 -- Informix locking problems

Path: emory!wupost!sdd.hp.com!hplabs!pyramid!infmx!aland From: aland@informix.com (Colonel Panic) Newsgroups: comp.databases.informix Keywords: locking problems Message-ID: <1991Oct12.221353.10005@informix.com> Date: 12 Oct 91 22:13:53 GMT References: <1991Oct11.211736.21158@mks.com> Sender: news@informix.com (Usenet News) Organization: International Brotherhood of Geeks, Local 619 In article <1991Oct11.211736.21158@mks.com> john@mks.com (John Postma) writes: >Has anyone out there run into problems with Informix only allowing 64 >row level locks between a begin and commit, causing a run-time error >on the 65th row. We are running Informix .03 SE, using SUN OS 4.03 and >the problem occurs in a commercial package (FOURGEN Accounting). I have >come across the problem with in-house code, but usually have implemented >some type of work-around by locking the entire table to be used. Our port >of Informix uses the .lok files for locking. If there is anyone out there >who has run into a similar problem, what have you done to alleviate the >problem?? Any suggestions would be greatly appreciated.... "Create locking" platforms, e.g. those with .lok files to track record locks, limit you to 64 simultaneous locks on each table. All SunOS platforms used create locking due to performance and integrity problems found in testing fcntl locking on older versions of SunOS. With SunOS 4.1.1, the problems don't seem to occur anymore, so we now use fcntl/flock locking on current Sun ports. This is effective with the 4.10.UC2 releases (and higher) of SE and C-ISAM on Sparc platforms and with 4.10.UE1 (and higher) on Motorola (when complete). If you upgrade, read the release notes carefully on how to make the locking conversion transparent using the RESETLOCK environment variable. (and if that note ain't clear, guess who you get to blame... |-:) I have no clue if FourGen has any dependencies that rule out upgrading to 4.1, so ask them first... X.10.03 tools can be used with 4.1 engines, but they won't be able to use new 4.X datatypes. >Thanks in advance, >John Postma. -- Alan Denney aland@informix.com {pyramid|uunet}!infmx!aland "The first computers were big clumsy machines that used vacuum tubes. By today's standards, they were extremely primitive. For example, they believed the sun was carried across the sky on the back of a giant turtle." -- D. Barry