[Exim] Exim using secondary mx despite primary being availab…

We are running exim version 3.22 on a Red Hat Linux 7.1 box and are
experiencing problems when mail is relayed to one particular domain. This
domain has two mx records one with priority 0, the other with priority 10.
Up until yesterday, mail was being relayed using lookuphost to the mx 0
server. At some point though this server failed and we got a 'Connection
refused' line in the exim log. The message that the mx 0 server couldn't
accept was successfully delivered to the mx 10 backup server.

From that point on, all mail destined for that domain was sent to the mx 10
backup despite the main mx 0 server having been restarted and accepting
mail. To try and get exim to use the mx 0 server again, I tried restarting
it and restarting the caching dns server it is using for resolution. Neither
of these actions made any difference. New mail for relay would still be
delivered to the mx 10 server. I have now got exim to deliver again to the
mx 0. I temporarily added a firewall rule to prevent access to the mx 10 and
then sent a mail to the domain. Exim failed to deliver to the mx 10 and
then tried the mx 0 (on the next queue run). I have now removed the firewall
rule and exim is successfully delivering to the mx 0.

Is this normal behaviour for Exim or this version of Exim? If it is specific
to 3.22, are later versions any different? Are there any options to control
retrying higher priority mail servers?

Thanks,

Phil