memory alloc and probe

[Glenn Slaven]

Are the probe thresholds and counts configurable?

> On 19 Jun 2014, at 4:07 pm, "MAGNIEN, Thierry" <thierry.magnien at sfr.com> wrote:
> 
> Hi,
> 
> 1) Tou have to check if you have objects that are removed from cache because of lack of space. They are marked  as "n_lru_nuked" in varnishstat.
> 
> 2) With your probe configuration, varnish will have to get at least 8 good results out of 10 from probes in order to consider the backend healthy. With an interval of 30s, this can take quite long ;-)
> 
> Regards,
> Thierry
> 
> 
> 
> -----------------------------
> De : varnish-misc-bounces+thierry.magnien=sfr.com at varnish-cache.org [mailto:varnish-misc-bounces+thierry.magnien=sfr.com at varnish-cache.org] De la part de Hernán Marsili
> Envoyé : mercredi 18 juin 2014 19:15
> À : varnish-misc
> Objet : memory alloc and probe
> 
> Hi,
> 
> We have two problems while using Varnish:
> 
> 1) we actually don't know how to determine if the amount of memory we assigned to the MALLOC is enough or not. varnishstat -1 show this:
>  
> SMA.s0.c_req               58035        34.08 Allocator requests
> SMA.s0.c_fail                  0         0.00 Allocator failures
> SMA.s0.c_bytes        4830340101   2836371.17 Bytes allocated
> SMA.s0.c_freed        2864339421   1681937.42 Bytes freed
> SMA.s0.g_alloc             44952          .   Allocations outstanding
> SMA.s0.g_bytes        1966000680          .   Bytes outstanding
> SMA.s0.g_space         181482968          .   Bytes available
> SMA.Transient.c_req        72960        42.84 Allocator requests
> SMA.Transient.c_fail           0         0.00 Allocator failures
> SMA.Transient.c_bytes   4366160011   2563805.06 Bytes allocated
> SMA.Transient.c_freed   4365783718   2563584.10 Bytes freed
> SMA.Transient.g_alloc          350          .   Allocations outstanding
> SMA.Transient.g_bytes       376293          .   Bytes outstanding
> SMA.Transient.g_space            0          .   Bytes available
> 2) PROBE: we configured the probes as follows. What we noticed is, once the backend is mark as sick, even when the service is restored, Varnish continues to answering with a GURU MEDITATION. Is like VARNISH never checks back again the backend to see if it is healthy again. Is that a misconfiguration we have?
> 
>  .probe = {
>       .url = "http://www.tycsports.com/balancer/alive1.html";
>       .interval = 30s;   #chequear cada X seg
>       .timeout = 10s;    #timeout de la prueba
>       .window = 10;      #el probing funciona como sliding window
>       .threshold = 8;   # cuantos del window tienen que estar ok para declararl$
>  
> Any help will be much appreciated! tnx :)
> 
> Regards,
> Hernán.
> _______________________________________________
> varnish-misc mailing list
> varnish-misc at varnish-cache.org
> https://www.varnish-cache.org/lists/mailman/listinfo/varnish-misc