Ontem ao sair

do trabalho passei no continente para fazer algumas compras, coisa pouca, mas necessária para poder variar um pouco a gastronomia do fim do dia.
Ao chegar à caixa ponho as minhas coisas em cima do tapete e aguardo a minha vez. De repente a minha atenção recai nas compras do cliente da frente que passo a descrever: gomas, donuts, batatas fritas, nougat’s, bolachas, filipinos e chocolates. Sem sequer olhar para a pessoa imaginei..isto…devem ser adolescentes às compras ou então universitários..eheh.

Olhei e confirmei…eram os primeiros..ah a libertinagem calórica e a inconsciência juntas num fervoroso festim de coisas realmente nutritivas…ou não!!

Claro está que os meus items – couve-rouxa, alface, pão, peitos de frango e pacotinho de vinho branco para tempero…em comparação não tinham o mesmo glamour.

Mas pronto e daí também quando me perguntaram se tinha o cartão do cliente eu tive a “adultez” de não responder…e cito “a minha mãe tem disso lá em casa ..eu não”. Tão crescido que sou…eheh

E depois dizem..

que os informáticos são isto e aquilo, não prestam, são geek’s, só pensam em gajas e tal.

Situação do dia: Após queixas de problemas com emails, vim a descobrir que uma conta entrou em loop precisamente quando estava a enviar um email de 5 MB. Até aqui ok, teria solução, não é inédito dado que qd acontece normalmente o destinatário queixa-se pouco depois uma vez que começa a receber o mesmo email vezes sem conta.

O caso em particular abrangia um leque de cerca de 15 emails de destino, e portanto para além da repetição de emails estaria em causa a própria qualidade da rede, por saturação.

Mas não, ninguem se queixou. Coincidência..ou propositado?! E porquê propositado…pq o mail em causa era nada mais nada menos do que o típico conteúdo de “gajas”.

Aterraram o circuito, aterraram o servidor, mas se existe alguma satisfação é pensar como é que o originador se sentiu quando a minha colega(já uma sra. mãe, que se diga a bem da verdade ) lhe telefonou para lhe dar assistência a tirar o loop de um email…condimentado!

Os outros…suponho que não se queixaram precisamente pelo conteúdo. Ou então foi uma coincidência…..NOT.

enfim, claramente uma situação em que os informáticos foram os bons da fita desta vez

It’s mid-night..

and the sleep is lurking as I write this post. Disturbed is in the background to rock this cradle. These past few days have been a bit stressful. My biggest dark cloud over me has been the sensation of lack of closure in my projects. Not that they don’t get done, but I always keep getting the feeling that they could be better.
Learned how to reset the root pass on Mysql, how to integrate Mcafee’s Virusscan with Samba, messed around a bit with qmailroaster (a opensource qmail bundle of products to help manage a qmail gateway). Obviously these are not great achievements, and can easily be “googled” , but still  it’s nice to actually put them into practice to accomplish a real need.

Well, the lurking has ended..I’m officially ready to hibernate within the next 5 minutes. bye

A proxy with an identity crisis

Yesterday I was going to post this one, but I fell asleep..

It’s 1:30 am., I’ve just arrived home from a late intervention on a squid proxy. The idea was to change its position in the network’s topology, as well as use the downtime window and change the system’s OS version.

After some physical modifications on the cabling and vlan’s, I started redoing the proxy’s routing table and IP’s configuration. This one had a heartbeat daemon that changed the main IP between two identical systems(a prodution and a spare). Because I want to change the OS in both systems, I have to break this action into at least two phases. One for each, but maintaining a proxy always on-line.
As I carried the operations, I noticed that every “son-proxy” was complaining to classify the parent proxy as “DEAD parent”. A DEAD parent cache state is induced by the proxy’s control messages (icp messages, which is udp based). Bad conectivity can cause this. At first, and assuming that the firewall I had setup wouldn’t block that service, I did not found any reason why the link would be a bad one.

I was puzzled, and given that the problem would be potentially time consuming we went to dinner. After I came back, I started doing on tcpdump in their “sorry asses”, so I could understand a bit more why did this break. After some digging I came to a possible cause. The proxy service is linked on a moving ip (between both machines), so what I noticed was that despite both alias and real IP were on, I was having a cross-linked dialog between the parent and the son proxy. Basically the son would send a packet to the aliased IP, but the response came sourced from the real one, leading the son-proxy to conclude that the parent cache would be dead.

I fixed it, with a simple workaround, but my idea is to reassemble the heartbeat daemon after the platform’s OS is redone. I pondered on defying myself a bit more and go for a minimal cluster setup, using both systems, but for doing that I would need an external storage. And that would make the solution too much expensive for the results needed.

I will be looking to squid-guard as a potential add-on to the new platform..look’s promissing.

Fast-geeky post

Fast and brief tech post:

LVM..nice thing to learn;

milter-length for sendmail..not quite there yet;

spamassassin…trivial;

dynamic dns for now just a test-drive to get the feeling;

proxy cluster…huuuh will get there hopefully;

Samba with AD integration using LDAP technique..Still to determine if slowness is in the machine or in the technique;

And of course my favorite until now.. 10 minutes of pure rush.. deployment of a fully production image using image taken with mkcdrec (although beware of redundant partition labels in a dual disk sata environment).

Centos, HP alarmistic suite, and many..many..many ssh sessions ;

And yes..the probability that was almost certain…got confirmed….some windows server things.

As a conclusion I can say…so far so good…what’s next?