Hello > > Today we had found that our BDII fails the freshness test with the > usual LDAP modify error "Unknown error(28)". It turned out that the > size of the tmpfs partition for /var/run/bdii/db was too small (1.5G): > after some operation time, the DB grows larger, so there will really > be update errors that will lead to the non-fresh information in the > BDII. > I had created the ticket, > https://ggus.eu/ws/ticket_info.php?ticket=79393 > that enables one to customize the size for tmpfs partition and we're > currently running with 2.4G limit. > I wonder if any other sites who > run top-level BDII nodes and using tmpfs for database see this problems > or it is our local configuration strangeness that makes DB to grow > past the default limit? We had this problem on our top-BDIIs in NGI_PL but it has been fixed in December by increasing tmpfs size to 3.5GB. We also define NETWORK_TIMEOUT 10 in /etc/openldap/ldap.conf to prevent infoprovider from hanging on ldap queries if remote ldap port is filtered with -j DROP Currently we are facing timeouts but cause has not yet been identified 2012-02-19 13:00:03,929: [ERROR] Timed out while reading /var/lib/bdii/gip/provider/glite-info-provider-top 2012-02-19 16:11:22,056: [ERROR] Timed out while reading /var/lib/bdii/gip/provider/glite-info-provider-top 2012-02-19 18:04:57,927: [ERROR] Timed out while reading /var/lib/bdii/gip/provider/glite-info-provider-top Cheers, -- Lukasz Flis